energyaustralia · 3.3 mmne customer acquisition costs 27 3.4 mmne retail costs 28 3.5 mmne retail...

E N E R G Y A U S T R A L I A

RESP ON SE TO F RONTI ER ’S DRAF T REPOR T

FEBRUARY 2007

This page has been left blank intentionally

1 EXECUTIVE SUMMARY 2

2 ENERGY COSTS 5

2.1 Introduction 5 2.2 The Business of Electricity Retailing 5 2.3 Methodology 9 2.4 Efficient generation costs (LRMC) 10 2.5 Market-based approach 10 2.6 Greenhouse cost allowance 22 2.7 NEMMCO-related costs 23

3 RETAIL COSTS AND RETAIL MARGIN 24

3.1 Definition of a MMNE 24 3.2 Relationship between cost allowances 27 3.3 MMNE customer acquisition costs 27 3.4 MMNE retail costs 28 3.5 MMNE retail margin 37

4 APPENDIX A – KPMG REPORT FOR ENERGYAUSTRALIA ON

BENCHMARKING RETAIL OPERATING COSTS AND MARGINS

5 APPENDIX B – ADDENDUM TO KPMG REPORT FOR ENERGYAUSTRALIA

ON NET MARGIN USING COST BUILD-UP APPROACH

6 APPENDIX C – MMA REPORT ON ALLOWANCE FOR WHOLESALE COSTS

IN RETAIL TARIFFS

R e s p o n s e t o F r o n t i e r ’ s D r a f t R e p o r t

1

1 E X E C U T I V E S U M M A R Y

EnergyAustralia welcomes the opportunity to comment on Frontier Economics’ reports on energy costs, retail opex and retail margin for the Independent Pricing and Regulatory Tribunal. These are critical cost components in the setting of regulated prices and, as such, should receive due consideration.

In particular, wholesale energy purchases represent a significant cost to our retail business. It is important for the Tribunal to understand and appreciate the volatility inherent is these energy costs as we move to an environment where the ETEF scheme is no longer available.

EnergyAustralia highlights the Ministerial requirement for IPART to ensure that regulated prices reflect the cost and risks of retailing electricity. Key to achieving this objective is setting appropriate allowances for each of the key cost inputs. With this in mind, EnergyAustralia focuses its response on four key areas of interest:

Long Run Marginal Cost (LRMC)

EnergyAustralia believes LRMC is a useful reference in the absence of an effective, well-functioning market. For this reason, LRMC is appropriate for assessing the compliance cost of the GGAS and MRET schemes where the spot markets lack sufficient capacity to accommodate large spot purchases. LRMC also more appropriately reflects the way that the bulk of green compliance has been transacted, through long term deals underwriting the construction of new plant.

EnergyAustralia support Frontier’s view that in the case of energy purchases LRMC may be useful “to assist IPART in judging the most appropriate level of market costs”. However we stress that LRMC itself does not reflect the medium term exposures faced by a retail business, which are more appropriately reflected in a market based hedge analysis.

Hedge costs

Hedge costs reflect the reality of purchasing the bulk of energy on behalf of our customers. The difficulty for the Tribunal is that they are required to set a regulated price up-front for the next three years, yet appreciate that pool prices, cost of hedging instruments and load will change from day to day. The resulting volatility and risk cannot be understated. The body of our response goes to some length to explain the extent of this volatility and its implied risk.


2

EnergyAustralia appreciates Frontier’s efficient portfolio approach for what it is – a calculation on the way a retailer could ‘theoretically’ hedge its regulated load from quarter to quarter. Although technically ‘efficient’, given the inputs provided, we found that the portfolio construction assumed presents a fundamental and significant departure from the manner in which EnergyAustralia, and arguably any other prudent retailer, would hedge its portfolios.

To facilitate a better understanding of the true cost of supply, and the degree to which it may vary, EnergyAustralia propose that Frontier Economics explore the representation of its hedge cost analysis to reflect an appropriate degree of confidence that the energy cost allowance is commensurate with the inherent costs and risks the hedging the regulated load profiles consistent with the efficient frontiers, as shown below.

50 % Confidence

84 % Confidence

97.5 % Confidence

The energy cost differential between EnergyAustralia, Integral Energy and Country Energy in Frontier’s draft report is not consistent with our understanding of the real cost to supply these areas. We suspect this may be a function of the input load data used in Frontier’s analysis – an issue that should be resolved before finalising the report.

The allowance for NEMMCO fees in Frontier’s draft report is consistent with our understanding of our future obligations and as such represents an appropriate amount should IPART set an ex-ante allowance.

Retail costs

EnergyAustralia believes an allowance of $70 per customer per annum for operating costs is insufficient. The allowance recommended by Frontier was based on the historical costs of standard retailers, with no regard to future expected cost increases. Although EnergyAustralia does not, in principle, disagree with using standard retailers as a proxy for a stand-alone MMNE retailer, we believe doing so needs to take account of some inherent differences between the two. Frontier


3

acknowledges some of these differences but do not attempt to quantify or address them.

We also believe Frontier need to refine its approach to benchmarking. International benchmarks need to be referenced correctly and Australian benchmarks should be the most recent. When benchmarked effectively, it is visibly evident the Frontier allowance for operating costs is below that provided for in other regulatory decisions.

Frontier’s recommended allowance for customer acquisition costs (CAC) is broadly in line with EnergyAustralia’s recommendation. We note the annual CAC allowance is sensitive to customer retention / churn assumptions which in turn reflect the level of competition in the market. As a result, we seek consistency between IPART’s forward expectations of competition and the annualised value of CAC allowed.

Retail margins

Frontier has taken a ‘three-pronged’ approach to calculating retail margin, including the use of the expected returns approach. EnergyAustralia recognise that in a regulatory setting, the expected returns approach is somewhat unorthodox. Notwithstanding, in the context of this retail price review, the expected returns approach is superior to the bottom-up approach. The latter is more appropriately suited to an asset-focussed business, rather than one whose value lies predominantly in its intangible assets (ie. customer base). To the extent this approach has captured systematic risks EnergyAustralia are satisfied that this mechanism appears to deliver an appropriate return. Of far greater concern are the significant non-systematic risks that have been excluded from this process. We urge and understand that these will be considered in setting the energy cost allowance.

The expected returns approach should be supported by benchmarking to ensure it delivers a sensible outcome. Frontier should also have regard for available market data. The market evidence is likely to provide the best available indication of the sustainable margin a MMNE might reasonably expect to earn, which might justify the decision to enter a market. Relevant market evidence might include independent experts’ reports and broker reports. Expert advice from our consultants, KPMG, suggests that a retail margin of between 5% and 8% is not unreasonable when referencing market-based data.


4

2 E N E R G Y C O S T S

2 .1 In t roduc t ion

The business of retailing electricity is unique in relation to the amount of risk faced by the businesses and the effectiveness of the tools available to manage this risk. It is critical that this environment is well understood and reflected in an appropriate energy cost allowance.

We have observed a tightening supply/demand balance in NSW over time and this is starting to manifest itself in the forward contract market. It is crucial that the energy cost allowance be set at a level that will encourage investment in the NSW generation sector. The Californian electricity market was characterised by volatile spot markets, fixed retail tariffs and inadequate hedging facilities for retailers prior to its spectacular collapse. The adequacy of hedging instruments is related to the level of risk margin allowed in the tariffs, the smaller the allowance the smaller the amount of uncertainty a retail business can manage without risking insolvency. A key aspect of this next Determination is the removal of the ETEF which currently provides a near perfect hedge to standard retailers. The removal of the ETEF introduces significant risks to the standard retailers. It is therefore vital that the Determination ensures appropriate costs and risks are reflected in setting the regulated tariff.

Given the plethora of significant risks that this industry already faces it would be prudent to allow for significant regulatory changes as a pass-through provision in the Determination. Events such as the possible introduction of the NSW Renewable Energy Target [NRET], a change in NEM region boundary definitions or changes to the level of VoLL are examples of potential events that may significantly impact the cost of retailing electricity. EnergyAustralia believes the pass-through mechanism in IPART’s existing NSW Distribution Pricing Determination is an effective tool to manage and account for such unforeseen or uncontrollable costs. This has been demonstrated with the introduction of the Guaranteed Customer Serviced Standard [GCSS], new reliability standards and the Energy Savings Fund.

2 .2 The Bus iness o f E lec t r i c i t y Re ta i l i ng

The business of retailing electricity is heavily dependant on the risk management of energy purchases which represent around 70% of a retailers cost. There is, however, significant uncertainty surrounding the actual outturn cost of these purchases which sets electricity retailing apart from any other industry.


5

2.2.1 The Product

To begin, consider a basic stand-alone retailer which does not ‘produce’ any goods, rather they must purchase everything they sell. In the case of regulated electricity customers the retailer also promises to provide an unknown quantity of electricity at a fixed price. In the industry this is referred to as a ‘whole-of-meter’ swap [WoM]. If retailers could purchase such products, this business would look much like any other retailing business and would attract the attention of Australia’s retailing giants. The only WoM swap available that perfectly hedges the regulated customer load profile is the ETEF. This product exists through regulation and is being removed over the course of this next Determination and is unavailable to a MMNE retailer.

2.2.2 The Spot Price

Retailers cannot purchase the product they sell. Although the physical delivery of electricity is generally sufficient to meet its customers’ requirements when it is demanded, retailers are obliged to purchase this energy on behalf of their customers from the spot market at the spot price. Therefore supplying the agreed [unknown] volume to the customer is generally low risk, however managing the cost of this purchase is exceptionally difficult as the spot market is extremely volatile. For the past four financial years, from July 2002, average time-weighted NSW spot prices have ranged from $32.37/MWh to $39.33/MWh pa. This represents a variation of 21% from the cheapest year [FY03/04] to the most expensive year [FY04/05].

Extreme price events contribute significantly to the variation in average prices. With average spot prices around $30-40/MWh, we might consider half-hour prices of ten times this [1000% above average] to be extreme. Historically prices above $300/MWh have occurred less than 0.5% of the time in any given year [<50hrs/yr] in NSW. However these events contribute a massive 25-50% of the average spot price for the year. This is because half-hour prices can range up to $10,000/MWh, more than 28,000%, above the average level.

Graph 11 below highlights the impact of extreme price events on the volatility of wholesale portfolios for regulated customers. The starting point for this chart is average NSW spot prices excluding extreme prices [above $300/MWh]. For the four financial years from July 2002 these prices ranged from $24.32/MWh to $26.80/MWh, representing a variation of 10% from the lowest year [FY03/04] to the highest year [FY04/05]. All other prices are measured relative to this position. The solid black line (B) indicates the premium that extreme spot prices have contributed to average spot prices each financial year. In 2004/05 the average spot price

1 The data is sourced from NEMMCO and reflects spot prices at the NSW regional reference node [NSW RRN]. The period spans four financial years from July 2002 to June 2006. This period captures the introduction of FRC [post Jan 2002] in NSW and Victoria and also the increase of the market price cap (VoLL) to $10,000/MWh [post April 2002].


6

excluding extreme events was $26.80/MWh, but increases by $13.53/MWh to $39.33/MWh - a contribution of almost 50%.

Graph 1 - Relative spot costs and impact of extreme prices

2.2.3 Combining Spot Price Volatility and Load Profiles

Exacerbating the volatility introduced by extreme spot price events is its correlation with the load profile that retailers are buying from the spot market. The regulated load profiles are positively correlated with spot price movements and amplify this volatility. Electricity is practically non-storable requiring retailers to buy the energy consumed by their customers at the time they use it. Demand for electricity is heavily dependant on weather conditions as well as the day of the week, time of day and general seasonal patterns.

The dashed coloured lines (C) on the chart above indicate the premium of the regulated load profiles to average spot prices when extreme price events are excluded. The spot cost of these load profiles represents a relatively stable [10% +/-5%] premium to average spot prices (excluding extreme prices). In contrast the solid coloured lines (A) show the relative spot cost of the regulated load profiles when extreme price events are included. This highlights the impact of the correlation between regulated load profiles and extreme price events. For the Country Energy regulated load profile the impact of extreme prices adds 45-70% to the underlying cost, 50-75% for the EnergyAustralia regulated profile and 65-110% for Integral Energy’s regulated profile.


7

2.2.4 Hedging Mismatch and Asymmetry

Retailers must also deal with the fact that it is not possible to perfectly hedge this price uncertainty and that any hedges one does purchase are bought well in advance, typically years, and well ahead of any valuable knowledge of the daily weather patterns.

Because the load and spot prices are correlated, hedging with imperfect products tends to increase the expected [50% confidence interval] cost when compared against historical spot prices, but is intended to reduce the potential for uncertainty in the portfolio outcomes. However, contracting levels and products also affect the ensuing spot price. Therefore, as noted by Frontier2, the true opportunity cost of hedging should be considered against an assessment of “spot price in the absence of the contract (as distinct from the observed difference between contract and spot prices once the contract has been agreed)”.

Hedging tends to increase the apparent expected cost of the portfolio as carrying more hedge cover than spot purchases will have a suppressing effect on spot prices as those contracts are ‘bid’ into the spot market. Conversely, as extreme weather and prices occur, load and spot price tend to increase. Should load increase such that one is buying more volume from spot than one has contracts for, this will result in a short contract position leaving a portion of one’s spot purchases unhedged. As seen previously, while these extreme events occur less than 50 hours per annum they have an enormous impact on the cost of the wholesale energy purchases.

2.2.5 Risk Containment

The longevity of a retail business is dependant on its ability to contain the uncertainty of its costs. Unlike most other businesses, 70% of an electricity retailer’s costs are exposed to significant uncertainty. Furthermore, these risks are focused within a small number of discrete but extreme pricing events across the year making them expensive and difficult to manage, and too significant to ignore.

Because the potential adverse variation in expected wholesale costs is so significant, the businesses cannot tolerate extended occurrences of these events. Therefore, hedging portfolios must be sufficiently robust to withstand most plausible outcomes. Recognising this, EnergyAustralia’s Earnings-at-Risk trading limits risk enforce a 97.5% confidence interval using a monte-carlo based analysis. The monte-carlo simulations consist of 10,000 unique half-hourly spot price simulations, each spanning 5 years. It is worth noting that these distributions are bi-normal, indicating that while most of the spot prices in any one simulation exist at the low end, there is

2 Frontier Economics, Energy Costs Draft Report – December 2006, Section 5.2.


8

also a number of extreme price outcomes. It is therefore important to capture the existence of extreme prices and their correlation with load.

The foregoing analysis discussion is predicated on perfect historical knowledge. To complete the picture, retailers must purchase their hedge positions months and years in advance. At this time there is substantial uncertainty regarding the magnitude and shape of the load to be hedged, uncertainty around the final spot price outcome and uncertainty over movement in contract prices. Most significantly, there is almost no information in this time-frame about daily weather conditions or extreme price events in the spot market. Although short-lived, these extreme conditions significantly impact the cost of energy purchases.

As it is impossible to know all or any of this information with any degree of certainty, the business of energy retailing is predicated on creating a robust portfolio that can withstand most plausible outcomes. As a going concern the retail business must avoid carrying a level of risk that may lead to insolvency. Should this occur to a major player in the NEM it would have cash-flow consequences for the whole market.

Finally, extend this analogy beyond a stand-alone retailer to a vertically integrated generator-retailer. It may be argued that the ‘gen-tailer’ is a less risky model. In theory a generation asset is perfectly anti-correlated to a retail portfolio. However, while vertical integration may reduce the expected cost of a retail portfolio it comes with its own risks, and is in fact more risky than owning a firm hedge. Generators are subject to physical risks such as planned and forced outages, transmission outages, de-rating in hot weather and water or fuel constraints to name a few. Moreover, because so much value is attributable to the extreme price events, there is significant risk the generator will not be operating when these events occur and therefore be of little value. All of these risks are factored into the cost of hedge contracts offered by generators which is one reason why firm contracts trade at a premium to spot market prices. To this extent, vertical integration actually increases market risk [standard deviation] although it should reduce the expected costs [mean] as there is one less firm to support and there are significant synergies potentially optimising green, gas and electricity portfolios.

2 .3 Me thodo l og y

A strong theme throughout Frontier’s draft report on retail margin and retail opex is the prudence of conducting a reasonableness test to ensure that the values determined using different methodologies (which require a degree of professional judgment) lead to a reasonable and sensible outcome.

Determining an appropriate allowance for energy costs also requires a degree of professional judgement. The care taken in assessing reasonableness evident in Frontier’s draft report on opex and margin appears absent in its report on energy costs.


9

Without having access to Frontier’s detailed modelling, EnergyAustralia has conducted its own reasonableness assessment. It strikes EnergyAustralia as unusual that the LRMC energy cost for EnergyAustralia is around $5/MWh less than the mid-point average of the IES report used in the current retail price Determination while input costs have generally increased since the IES review. Also the cost differential between EnergyAustralia, Integral Energy and Country Energy in Frontier’s draft report is materially inconsistent with our understanding of the real cost to supply these areas.

2 .4 E f f i c ien t genera t ion cos ts (LRMC)

EnergyAustralia supports Frontier’s recommendation that LRMC be used as an indication when considering the hedge costs analysis, but do not support LRMC as the primary basis for assessing energy costs as it does not reflect the reality of hedging. EnergyAustralia also support Frontier’s recommendation that Stand Alone LRMC form the basis of the Green costs allowances.

Similar to its hedge cost analysis, Frontier’s LRMC results show a real decline in energy prices year-on-year with no explanation why. Assuming all assumptions are correct, then Frontier’s LRMC estimates represents a 50% probability of coverage.

EnergyAustralia commissioned McLennan Magasanik Associates (MMA) to prepare a report to address the relevant energy cost allowances canvassed in the Minister’s Terms of Reference for the 2007-10 Regulated Retail Price Review. Using the same cost estimates assumed by ACIL Tasman in a report recently commissioned by NEMMCO, MMA derived an LRMC value of $57.70/MWh with an 80% probability of coverage in June 2006 dollars. MMA’s report is included in Appendix C of this response.

2 .5 Marke t -b ased approach

As discussed in Section 2.2, the business of retailing electricity is predicated on establishing a robust wholesale energy portfolio that can withstand most plausible outcomes. At the time contracts are purchased there is substantial uncertainty regarding the magnitude and shape of the load to be hedged, the final spot price and movement in contract prices. Most significantly there is almost no information in this time-frame about daily weather conditions or extreme price events in the spot market. Although short-lived these extreme conditions significantly impact the cost of energy purchases.


1 0

Recall that these risks exist in the retail business because:

•

•

•

retailers provide a fixed price variable volume [WoM] swap to customers;

retailers purchase this variable volume at a variable [volatile] spot price; and

the tools for hedging this price uncertainty are imperfect.

The residual risks are so large that inadequate coverage compromises the long term ability of the retail business to operate as a going concern. Therefore a retailer’s wholesale portfolio must be sufficiently robust to withstand a wide range of possibilities yet still be sufficiently cost effective to allow the business to compete and remain commercially viable.

2.5.1 Model Inputs

Spot Forecasts

We note that each standard retailer provided a spot price forecast in the Information Request and these were used as inputs to Frontier’s modelling. There was, however, only one such forecast from each standard retailer and the format of the forecast was a single value per calendar quarter and therefore absent of any information about half-hour spot price volatility, which is paramount to evaluating an optimal [robust] hedge portfolio. For reference, EnergyAustralia provided the following advice accompanying the spot price forecast:

[Forecast] NSW spot prices are calculated from the [forecast] swap prices with peak spot prices being peak swap prices less a risk premium of $3/MWh, and spot off-peak prices being equal to off-peak forward prices.

To ensure the optimal robust portfolio is identified through Frontier’s modelling process the spot forecasts must reflect the broad and extreme range of possibilities, not just those observed in recent history. The history we observe is a function of specific conditions at that time and so does not give a complete view of the possible risks. In particular, the spot prices are heavily dependant on the prevailing weather conditions and hedge positions at the time. While history is a useful tool from which to learn, it does not in itself contain the full picture. Consistent with the methodologies adopted in EnergyAustralia’s risk management framework, a prudent retailer would consider a monte-carlo simulation involving 5,000 to 10,000 spot price forecasts, each with 17,520 half-hourly observations.

Furthermore, we concur with Frontier’s assertion that the spot price forecasts should be on the basis of spot prices “in the absence of the contract”, in order for their model to appropriately assess the opportunity cost of hedging or not hedging. Such spot forecasts are difficult to determine from spot price history as this regulated load has always been effectively fully hedged with ETEF or vesting contracts.


1 1

Contract Prices

Standard retailers were also required to provide their view of forward contract prices as part of IPART’s Information Request. Frontier assessed the reasonableness of these forward curves against AFMA, ICAP and d-Cypha data. From Frontier’s draft report it seems that the forecasts were broadly consistent amongst themselves and were also shown to be consistent with the third-party sources.

However, it is well understood that the AFMA, ICAP and d-Cypha curves reflect relatively small [marginal] trading volumes compared to the volume required to replace the ETEF. Therefore it is clear that the forecasts provided by the standard retailers reflect the observable traded market for 5MW blocks of energy, which is known not to reflect the significant liquidity requirements of replacing the ETEF, which must be considered when assessing the cost of hedging the regulated load profile. Therefore these forecasts should be recast by the standard retailers to account for such issues as liquidity risk in Frontier’s modelling and final results.

Load Data

The load profile that a retailer is exposed to when winning a small customer is the Net System Load Profile [NSLP] and plus a proportion of Controlled Load Profile [CLP] if the customer also has off peak electric hot water. Each network area has a unique NSLP and CLP, and within that network area every customer with a Type 6 accumulation meter is deemed to follow the half-hourly profile of the relevant NSLP and/or CLP.

The regulated load which is under investigation in this review is, in principle, made up of small customers and so should ‘look’ like an appropriate combination of NSLP and CLP shapes. However the ETEF load shape is calculated somewhat differently to the NSLP and any underlying oddities in those calculations such as the treatment of embedded generation can have a significant impact on the resulting profile. Hence the resulting ETEF load shape held by NSW Treasury for settlement of the ETEF may appear different to the intrinsic shape which a MMNE retailer would be exposed to.


1 2

Frontier2003 2004 2005 2006 Forecasts

NSW TWA Spot Price 32.91$ 32.37$ 39.33$ 37.24$ EnergyAustralia Regulated Load Profile

NSLP - - - - -CLP - - - - -NSLP + CLP Mix - - - - -% CLP 14% 12% 13% 13% -

Integral Energy Regulated Load ProfileNSLP 9% 12% 18% 13% -CLP -6% 0% -1% 1% -NSLP + CLP Mix 6% 9% 15% 12% 3%% CLP 18% 16% 16% 14% 31%

Country Energy Regulated Load ProfileNSLP -8% -7% -6% -7% -CLP 9% 0% 4% 5% -NSLP + CLP Mix -7% -8% -7% -7% -13%% CLP 20% 16% 15% 14% 28%

Historical Spot Costs Relative to EnergyAustralia Profiles

Graph 2 - Comparison of Net system Load Profiles

Graph 2 shows the actual half-hour load profile of the Integral Energy, Country Energy and EnergyAustralia NSLP’s during a week of warm weather in February 2005. This example highlights the volatility of the Integral Energy NSLP compared to those of the EnergyAustralia and Country Energy network areas. Notably Integral Energy’s NSLP more than doubles its usual daytime peak demand from around 700MW to a peak of 1,800MW, a 160% increase. EnergyAustralia’s NSLP increases 60% and Country Energy’s NSLP moves 55% on this occasion.

Table 1 summarises the relative spot costs of the NSLP and CLP for the three network areas. Relative to the EnergyAustralia NSLP, Integral Energy’s NSLP cost has been 9-18% higher over the four years since July 2002. It is worth noting that the variation in this ratio appears to be positively correlated with average annual spot prices.

Table 1 - Relative Costs between Load Profiles


1 3

The regulated load profile should represent a sensible combination of NSLP and CLP and this is what we have used to compare to Frontier’s analysis. We know that the EnergyAustralia regulated customer base contains around 14% controlled load by volume, and this has been used to mix the EnergyAustralia NSLP and CLP. We have estimated the mix for Integral Energy and Country Energy at around 16% each. While there is a degree of estimation in mixing the NSLP and CLP, there is no estimation in the calculations of the component parts [ie. the NSLP and CLP]. Frontier could also remove this uncertainty as they will be privy to the actual mix of NSLP and CLP in Country Energy and Integral Energy areas.

When comparing the Integral Energy NSLP+CLP mix we observe a range of 6-15% premium to the EnergyAustralia NSLP+CLP mix. This should be directly comparable to the analysis of Frontier. Similarly the Country Energy NSLP+CLP mix has cost around a 7% discount to the EnergyAustralia NSLP+CLP. The right-hand column shows the comparable results from Frontier’s analysis indicating that the Integral NSLP+CLP is around a 3% premium to EnergyAustralia, while Country Energy’s profile is a 13% discount to EnergyAustralia’s. Based on this data, Frontier’s results imply that Country Energy’s regulated load profile contains approximately 28% Controlled Load by volume and that Integral Energy’s regulated load profile contains approximately 31% Controlled Load by volume. This appears incorrect and warrants further investigation.

2.5.2 Portfolios

Graph 3 and Graph 4 below illustrate the portfolio construction resulting from Frontier’s modelling. Observing the ‘conservative’ portfolio in Graph 3, it is interesting that such a large portion of the portfolio is covered in caps, especially in Q1. Moreover these portfolios are generally short to maximum demand by some 10%. Against the regulated load this might equate to a short position on an extreme day of some 100MW. If this were to occur on a hot summer afternoon [most likely], with spot price at VoLL for 6 hours from 12pm to 6pm, this would result in a loss of $6M per day.


1 4

Graph 4 - ‘Elbow’ portfolio construction

ad

turn increase the cost of the hedges and

change the spot outcomes in the market.

Graph 3 - ‘Conservative’ portfolio construction

The ‘elbow’ portfolios in Graph 4 are similar in length to the conservative portfolios but even more extreme in their structure from one quarter to the next. One can see that Q2 is nearly all caps, while Q3 is all swap. These portfolios represent a significant departure from the existing hedge positions in the market and hence render the use of any spot price history almost redundant in assessing them. Moreover they contain significant spot exposure both under the caps and in outright short positions. Given the bias toward extreme price events affecting spot costs, these portfolios are far from robust and indeed will move with spot movements.

These portfolios are also well outside the risk limits currently permitted at EnergyAustralia and indeed any prudent retail business operating in the NEM. It isalso unlikely that there will be sellers of such products in the volumes described. Caps tend to be sold by peaking generation while swaps tend to be sold by base loand intermediate plant. This proposed structure would require enormous peaking capacity in Q2, to be superseded by base load capacity in Q3 and would thus requiresubstantial redundant plant which would in

Conservative

-

0.50

1.00

1.50

2.00

2.50

3.00

Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2

2007 2008 2009 2010 2007 2008 2009 2010

PK PK PK PK PK PK PK PK PK PK PK PK OP OP OP OP OP OP OP OP OP OP OP OP

MW

(per

nor

mai

lsed

sha

pe)

CapSwapMaxAvg

Elbow

-

0.50

1.00

1.50

2.00

2.50

3.00

Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2

2007 2008 2009 2010 2007 2008 2009 2010

PK PK PK PK PK PK PK PK PK PK PK PK OP OP OP OP OP OP OP OP OP OP OP OP

MW

(per

nor

mai

lsed

sha

pe)

CapSwapMaxAvg


1 5

2.5.3 Efficient Frontiers

The efficient hedging frontiers presented by Frontier Economics contain a wealth of information and warrant further inspection. 5 shows the frontiers for hedging the EnergyAustralia regulated load profile for the financial year 2008/09.

Firstly, we observe that the model outputs [the four frontiers] seem to be quite sensitive to the associated spot price forecasts. We note from Frontier’s draft report that the contract forecasts from the standard retailers were broadly consistent with each other, while the spot price forecasts varied considerably. The resulting frontiers also vary in line with the individual spot price forecasts. Even with no explicit knowledge of the underlying portfolios, we can see that these frontiers imply a portfolio with significant exposure to spot prices, either under $300/MWh caps or as outright spot exposure. The fact that the four resulting frontiers are materially different indicates immediately that these ‘optimal’ portfolios are not robust across different spot price scenarios. These are therefore inappropriate as risk management tools for retailers as they largely mirror the spot price.

EA Reg Load 08/09

ETEF

Non Systematic Risks[excluded from margin]

Systematic Risks[Included in margin]

EA Reg Load 08/09

ETEF

Non Systematic Risks[excluded from margin]

Systematic Risks[Included in margin]

Graph 5 - Efficient Hedging Frontiers for EnergyAustralia's Regulated Load Profile

Secondly, we observe that as these portfolios are hedged, we would expect greater uniformity of the outcomes. However these charts confirm that there is still significant risk in these portfolios, highlighted by both the size of the standard deviation (x-axis) and the variance across the four scenarios (y-axis).

Thirdly, we see that the standard deviation of each these scenarios also varies by around $1/MWh. This indicates uncertainty within the uncertainty. If we assume the distributions were normally distributed, standard deviations in the order of $2.50/MWh to $3.50/MWh would indicate that the portfolio cost could move by up to +/- $7.50 to $10.50/MWh. In fact the distributions are asymmetrical because the spot


1 6

price distributions are asymmetrical, which suggests there is greater risk of the portfolios costing more rather than less due to the existence of extreme price events.

It is worthwhile understanding the transition in risk management implied in these efficient frontiers. The existing WoM hedge the standard retailers rely on for regulated customers, known as the ETEF, represents zero risk in the context of these charts, and its price is set around $50/MWh in 2006/07dollars. With this risk profile IPART previously deemed it appropriate for the standard retailers to earn a 2% margin. However with the removal of the ETEF, standard retailers and MMNE are asked to manage significantly increased risks.

Table 2 is a summary of the results of the efficient frontiers for both the elbow and conservative points. Here we observe the year-on-year change in the range of -1.7% to -6.6%, a substantial real decline in energy purchase costs. Experience suggests (and supported by MMA) energy prices have been rising at around CPI-1% - in other words, a real decline of 1% per annum. It would seem the decline shown in Frontier’s results is well in excess of any historical evidence and would seem otherwise inappropriate.

Table 2- Summary of Efficient Frontier Results

2008 2009 2010 2008 2009 2010 2008 2009 2010EnergyAustralia Regulated Load Profile

Elbow Min 49.30$ 47.60$ 45.10$ - 1.70-$ 2.50-$ - -3.4% -5.3%Mean 51.05$ 49.70$ 47.90$ - 1.35-$ 1.80-$ - -2.6% -3.6%Max 52.80$ 51.80$ 50.70$ - 1.00-$ 1.10-$ - -1.9% -2.1%

Conservative Min 53.20$ 49.70$ 48.20$ - 3.50-$ 1.50-$ - -6.6% -3.0%Mean 54.70$ 52.45$ 50.80$ - 2.25-$ 1.65-$ - -4.1% -3.1%Max 56.20$ 55.20$ 53.40$ - 1.00-$ 1.80-$ - -1.8% -3.3%

Integral Energy Regulated Load ProfileElbow Min 50.70$ 49.40$ 46.80$ - 1.30-$ 2.60-$ - -2.6% -5.3%

Mean 52.35$ 51.25$ 49.20$ - 1.10-$ 2.05-$ - -2.1% -4.0%Max 54.00$ 53.10$ 51.60$ - 0.90-$ 1.50-$ - -1.7% -2.8%


Country Energy Regulated Load ProfileElbow Min 43.10$ 41.50$ 39.30$ - 1.60-$ 2.20-$ - -3.7% -5.3%

Mean 44.85$ 43.60$ 41.80$ - 1.25-$ 1.80-$ - -2.8% -4.1%Max 46.60$ 45.70$ 44.30$ - 0.90-$ 1.40-$ - -1.9% -3.1%


Frontier Modelling Results $/MWh Year on Year Change $/MWh Year on Year Change %

2.5.4 Non-systematic Risks

Section 2.5.3 discusses the introduction and magnitude of risks evident in moving from the ETEF hedge to a market-based solution. Frontier and SFG Consulting draw a distinction between systematic and non-systematic risks in their assessment of an appropriate retail margin. Unfortunately the risks identified as non-systematic are far greater than the systematic risks and have been specifically excluded from the margin allowance. We urge that these non-systematic risks must therefore be allowed for in the energy cost allowance. A review of the significant non-systematic risks is detailed below.


1 7

Contract Price Movement [Estimation Risk]

Graph 6 shows the movement in the flat swap contract price for NSW FY2007/08, the first year of the next Determination period. This contract has moved from a low of $34/MWh in July 2003 to its current high of over $42/MWh – an increase of 23%. The chart also highlights the volatility in this movement. A recent example of volatility was where prices rose to $41/MWh approaching Q4 2006 on the expectation of hot weather and extreme price events as had occurred the two previous years. When these conditions did not eventuate the price fell to a trough of $37/MWh. Extremely hot conditions and bushfires struck in January 2007, predominantly in Victoria and South Australia, and the NSW contract is consequently trading at a new high of $42/MWh.

This uncertainty would normally be managed by reflecting these movements in prices being offered to customers at the time, therefore the retailer is buying [from the market] and selling [to the customer] at the same market rate. However this mechanism is not available to the retailer for regulated customers as IPART will fix a price for the next three years for all regulated customers. This risk should be allowed for in the energy cost allowance.

FY07/08 swap, $34/MWh in Jul03

FY07/08 swap, over $42/MWh in Jan07As a result of Hot VIC/SA Q1, bushfires, Load

Shedding and VoLL events

FY07/08 swap, Rallied onexpectations of Hot Q4

FY07/08 swap, Collapsed with mild Q4 weather

FY07/08 swap, $34/MWh in Jul03

FY07/08 swap, over $42/MWh in Jan07As a result of Hot VIC/SA Q1, bushfires, Load

Shedding and VoLL events

FY07/08 swap, Rallied onexpectations of Hot Q4

FY07/08 swap, Collapsed with mild Q4 weather

Graph 6 - History of FY07/08 flat NSW swap contract

Hedge Mismatch

Hedge mismatch risk arises because the hedging tools are imperfect. It is exacerbated due to the existence of extreme price events and their correlation to


1 8

movements in regulated load. This tends to mean that the risk profile is asymmetrical hence when weather is mild and prices are low, load tends to be down and retailers generally have long contract positions. However when weather and prices are extreme load tends to be very high, often leaving the retailer short to some extent which can occur at very high prices. Both of these situations increase the cost of energy purchases. Having a 100MW [eg. 10% of the capacity that day] exposure for six hours during a hot summer afternoon from midday to 6pm when prices are $10,000/MWh would cost the retailer around $6M each time this occurs. This energy would normally cost around $30,000 per event.

Forecasting and Demand

The principle uncertainties affecting energy purchase and hedge requirements are the success of sales campaigns and daily weather outcomes. Hedge portfolios are purchased well in advance of the period being hedged, some 2 to 3 years, and well ahead of any useful knowledge of these variables. Weather has a significant impact on the cost of a portfolio as small customer consumption is dominated by the energy requirements for heating and cooling which also largely drives the extreme price events. The sales success will drive the average level of demand [ie. how many customers were won], while the weather on any given day will drive the resulting half-hourly load shape and extreme price activity. The NSLP can vary in day to day capacity requirements up to 160%, and are correlated to extreme spot prices.

Regulatory

Significant exposures result from changes to regulatory decisions. Examples of these include changes to the market price cap [VoLL], region boundary changes and changes to the relevant ‘green’ legislation. These risks can be significant and are best managed with the ability to re-open the Determination in a limited way to account for these changes. The potential introduction of the NSW Renewable Energy Target [NRET] is a perfect example. This mechanism of limited re-opening has worked well in the Distribution Pricing Determinations for addressing the Energy Savings Fund, the new reliability standards and Guaranteed Customer Service Standards. EnergyAustralia propose that a similar pass through mechanism be adopted for the Retail Price Determination in order to minimise any regulatory risk.

Liquidity

The total NSW ETEF represents around 25% of the total NSW state load, and therefore a significant volume to be contracted. Transacting such volumes affects the market price as there is insufficient depth at the apparent prices in the broker markets for these to be representative of hedging this volume. This is normally managed by buying incremental contracts continuously throughout the year at the same time retailers offer to customers. This mitigation technique is not available as a result of this Determination and consequently this risk should be recognised in the energy cost allowance.


1 9

Spot Price

The specific spot prices that materialise in any given half hour are unknown and uncertain, especially in the timeframe that hedging and sales take place. Spot prices can vary from -$1,000/MWh to +$10,000/MWh and tend to be correlated with extreme weather and demand conditions. This risk should be allowed for in the energy cost allowance.

2.5.5 Extending The Efficient Frontiers and Prudent Risk Management

Upon inspection the efficient frontiers contain a wealth of information about the residual uncertainty in these hedge portfolios, as discussed in earlier sections. Graph 7 shows the efficient frontiers for the EnergyAustralia load profile for financial year 2008/09. This graph highlights a $4/MWh difference in expected values across the four scenarios and a standard deviation for the conservative points of $2.50/MWh to $3.50/MWh.

EA Reg Load 08/09EA Reg Load 08/09

Graph 7 - Efficient Frontier for EnergyAustralia Regulated Load Profile financial year 2008/09

Looking more closely at the conservative points by way of example, we can infer the distributions shown in Graph 8. This is a stylised representation assuming a normal distribution. In fact the distributions have ‘long tails’ to the right due to the presence of extreme price events.

The four coloured lines represent the uncertainty of the conservative points in Graph 8, each with its own mean and standard deviation. The yellow chart represents the combined distribution of the previous four. It is important to understand that the ‘expected cost’ shown on the efficient frontier charts represents the mean of these distributions and that a significant range of outcomes [as indicated by the standard deviation on the x-axis] are possible. Therefore choosing the mean yields a 50% chance [or confidence] that the cost will be lower..


2 0

50% POE [50% CI]mean

16% POE [84% CI]Mean + 1 SD

2.5% POE [97.5% CI]Mean + 2 SD

50% POE [50% CI]mean

16% POE [84% CI]Mean + 1 SD

2.5% POE [97.5% CI]Mean + 2 SD

Graph 8 - Distribution for conservative portfolios

Extending this concept further by moving to higher levels of confidence, we can expect an 84% confidence interval by choosing a cost one standard deviation above the mean [approx $56/MWh] in this case and 97.5% confidence at two standard deviations [approximately $59.70/MWh]. EnergyAustralia’s internal risk policy requires a 97.5% confidence interval, due to the asymmetrical nature of the risks and the ability of retail businesses to withstand adverse outcomes. Graph 9 illustrates the application of 50%, 84% and 97.5% confidence intervals on the efficient frontier chart.

EA Reg Load 08/09

50 % Confidence

84 % Confidence

97.5 % Confidence

EA Reg Load 08/09

50 % Confidence

84 % Confidence

97.5 % Confidence

Graph 9 - Representation of the confidence intervals on the efficient frontier

The Terms of Reference require IPART to ensure that regulated tariffs under its pricing Determination reflect underlying costs and risks. Accordingly EnergyAustralia believes that the 50% confidence interval costs identified in Frontier’s original efficient frontiers do not achieve this. Adopting a 97.5% confidence interval gives the Tribunal far greater coverage of the potential cost outcomes and consequently better ensures the costs are reflected in the energy cost allowance.


2 1

2 .6 Greenho use cos t a l low ance

2.6.1 LRMC versus market-based valuation

As noted in our submission to IPART on its Issues Paper, we recommend the cost of RECs be valued using latest technology wind generation, and the cost of NGACs be valued using latest technology combined cycle gas turbine generation. In contrast, a market-based approach to determining the value of green certificates is inappropriate. Spot markets for green energy certificates have insufficient capacity to accommodate large quantities of spot purchases. The reality of complying with these greenhouse gas reduction schemes involves entering into long-term contracts to underwrite the construction of the required project. As such, the LRMC of the appropriate technical solution is the best proxy for the green cost allowances.

2.6.2 Renewable Energy Certificates

EnergyAustralia supports the LRMC-derived value of RECs in Frontier’s draft report. The LRMC-based REC value closely aligns with MMA’s view of the forward price of our obligations under the Mandatory Renewable Energy Target (MRET) scheme.

2.6.3 National Greenhouse Abatement Certificates

EnergyAustralia believes the LRMC-derived NGAC price found in Frontier’s draft report is understated. MMA’s analysis is summarised in Table 3 and indicates the forward NGAC price to be more than $1/MWh higher in the first two years, and around $0.70/MWh in the last year than suggested in Frontier’s draft report.

Specifically AFMA Environmental Products data would appear to use a higher market cost but the resolution of the chart on p36 of Frontier’s report makes it difficult to determine the exact differences. We prefer the use of AFMA data than selected Broker data. It is also not evident what percentage of sales are assumed to satisfy the retailer’s GGAS obligations each calendar year. When finalising its report, we request that Frontier advise the impact of this in Clause 9.2 of the “Greenhouse Gas Benchmark Rule (compliance) No1” in the calculation of the estimate of Total State Demand for calendar year 2007 through to calendar year 2010. We also request Frontier indicate the assumed pool coefficient used for calendar year 2007 through to calendar year 2010. We expect the figure will continue to rise for the next three years given it is a five year trailing average lagged two years.


2 2

Table 3 - Summary of Costs estimated in MMA report

2 .7 NEMMCO-re la ted cos ts

EnergyAustralia believes that the allowance for NEMMCO fees in Frontier’s draft report is consistent with our expectations of our future obligations and as such represents an appropriate amount should IPART set an ex-ante allowance. However, we reiterate our view that (where possible) uncontrollable retail costs should be passed through to customers. To the extent that the cost is:

Not subject to the influence of the retail businesses; •

•

•

Subject to some other from of review or oversight; and

Readily observable

it should, in principle, be suitable for pass through treatment in the regulatory framework. EnergyAustralia considers that NEMMCO pool fees fits into this category of costs and so should be included as a separate component or ‘factor’ in the price control formula.


2 3

3 R E T A I L C O S T S A N D R E T A I L M A R G I N

EnergyAustralia is broadly satisfied with the analysis conducted by Frontier Economics and SFG Consulting in relation to retail opex and margin. Their analysis was consistent with the methodology established upfront, which EnergyAustralia understood and supported.

In the remainder of Section 3, we will direct our attention to those aspects of Frontier’s preliminary work which may be further refined to achieve the most appropriate outcome.

3 .1 De f in i t ion o f a MMNE

The Minister’s Terms of Reference require IPART to have regard for “mass market new entrant retail costs” and “mass market new entrant retail margin”. The Terms of Reference provide little guidance as to the actual definition of a MMNE, except that it needs to be “of sufficient size to achieve economies of scale.”

EnergyAustralia appreciates the degree of discretion Frontier must exercise when attempting to define a MMNE. A variety of alternative definitions could be put forward that are individually consistent with the Terms of Reference but comparatively dissimilar.

EnergyAustralia sought the advice of KPMG in developing an appropriate cost allowance for retail opex and retail margin for the next Price Determination (see Appendix A.) In order to develop an allowance, KPMG needed to first define a MMNE. It found that it would be reasonable that the MMNE would have the following characteristics:


2 4

We recognise that a MMNE could be defined in a variety of other ways and still be consistent with the Terms of Reference. Frontier has developed an alternative MMNE definition that we believe is not unreasonable. Importantly, Frontier have sought to explore both the scale and scope of a MMNE’s operations which will have a critical effect on cost assumptions. EnergyAustralia will address the issue of scale and scope of operations in the following sub-sections.

3.1.1 Scale of a MMNE

Small, niche retailers in Australia

Frontier argues there is evidence of a long, flat average cost curve which suggests economies of scale are achieved with a relatively modest customer base.

EnergyAustralia believes this observation, if true, is likely a function of the various stages in the growth of a retailer. Each stage of growth culminates in exceeding ‘critical mass’ in the number of customers and the resulting need to re-invest in a customer information system (CIS) that can cope with the complexities of an increasingly diverse customer base. This is conceptually depicted in the following graph:


2 5

Number of customers

Aver

age

cost

s

re-invest

Niche Mass market

re-invest

The re-investment in CIS observed through growth in customer numbers does not (so easily, if at all) occur in reverse. With continued retailer of last resort responsibilities, limited short- to medium-term ability to scale back an existing CIS, and a customer base that diminishes with greater competition, EnergyAustralia expects that average costs for a standard retailer will only increase over time – observe the additional blue line in the following graph:

Number of customers

Aver

age

cost

s

re-invest

Niche Mass market

re-invest

To conclude that average costs are consistent across a range of operational scales, although consistent with observations, should not ignore the cost impact now facing standard retailers with significant sunk investment in CIS platforms that are not downward-scalable in the immediate, relevant future.

3.1.2 Scope of a MMNE

EnergyAustralia believes the achievement of complete economies of scope for a MMNE critically hinge on the extent of vertical integration. EnergyAustralia agrees with Frontier’s definition of a MMNE being a stand-alone retailer. KPMG also supported this definition in its analysis.

Frontier recognise that a stand-alone new entrant would not have achieved all economies of scope. There’s an important implication here: as Frontier have used the standard retailers costs as a proxy for the MMNE, a discrete adjustment needs to be made to account for the fact that standard retailers enjoy the synergies of a common (ring-fenced) customer information system with its distribution businesses. We explore this issue in further detail in Section 3.4.1 – under ‘Cost synergies’.


2 6

When discussing the risk management benefits of vertical integration, Frontier notes:

“because the returns to retailing and generation are negatively correlated…a retailer / generator faces lower risk than a stand-alone retailer. In short, there are benefits associated with diversification. These benefits would be reflected in the retail margin.” (p. 10 – Retail opex and margin)

Yet when discussion risk decomposition, Frontier claim:

“(i)t is only the…systematic [non-diversifiable] component that requires compensation via a return in the form of a retail margin.” (p. 42 – Retail opex and margin)

We see this as a conflict of statements that needs to be understood and resolved in the final report.4

3 .2 Re la t ionsh ip be tw een cos t a l low ances

EnergyAustralia agrees that it is imperative that retail costs, retail margin and energy cost are assessed as a collective to ensure that no one cost type is ‘double counted’ or ignored. In reviewing its report, we are confident that Frontier have not twice counted any cost allowance. However, as argued in Section 2.5.4, we believe that there are systematic risks arising from the suggested ‘risk-return’ positions on the efficient frontier that are not adequately compensated elsewhere in the framework. We explain this in further detail in Section 3.5.

3 .3 MMNE cu s tomer acqu is i t ion cos ts

Frontier’s recommended allowance for customer acquisition costs (CAC) is broadly in line with EnergyAustralia’s view of an appropriate allowance. In establishing the annual CAC allowance, Frontier note that three variables affect the recommendation:

•

•

•

Total cost of acquiring a customer;

Years a customer will be retained; and

Discount rate.

In relation to the second factor listed, EnergyAustralia believes the level of customer switching can be used to establish an approximate number of years a customer will remain with a retailer. For the purposes of this response, we believe that ‘switching’

4 The assertion that a gen-tailer faces lower risk is perhaps naïve – see our discussion in Section 2.2.5


2 7

refers to the adoption of a market contract, regardless of whether it is with the incumbent retailer or a new entrant5. This is an important distinction because it captures the underlying competitive pressures of retaining a customer and is often not publicly available in competitive market analysis.

EnergyAustralia believes Frontier need not look to mature overseas markets to gauge an appropriate retention expectancy. We suggest a more appropriate approach would be to assess the current levels of switching in NSW and then take a forward view on the level of competition and in turn customer churn rates.

AGL, in its 2006 Full Year Financial Results, noted that the current annualised churn rates in NSW electricity market were about 9%6. This equates to a customer retention life of around 11 years, broadly in line with the assumed life in Frontier’s draft report. However, IPART’s decision on the level of regulated prices (and hence available margin) for the 2007-10 Determination period will strongly influence the future degree of competition and in turn the level of switching in NSW.

We understand that IPART is currently reviewing the level and effectiveness of competition in the NSW market; an analysis that will be made available as part of its draft Decision. Given the annual CAC allowance is sensitive to customer retention / switching assumptions, we seek consistency between IPART’s forward expectations of competition and the annualised value of CAC allowed. In simple terms, if IPART expects the degree of competition (and hence switching rates) will increase, this should be factored into the conversion of Frontier’s overall recommended CAC allowance into an annualised value.

3 .4 MMNE re ta i l cos ts

In its draft report, Frontier developed a cost allowance using:

•

•

a cost build up approach; and

benchmarks from both regulatory audits and jurisdictional allowances.

Frontier’s methodology is not dissimilar to the approach taken by KPMG in its review:

There are two main ways to benchmark the operating costs a MMNE could expect to incur:

• undertake a bottom-up, cost build-up exercise; and

5 Ministerial Council on Energy Standing Committee of Officials, Phase Out of Retail Price Regulation for Electricity and Natural Gas - Draft Effective Competition Criteria Consultation Paper, July 2006, p. 9. Customer switching behaviour includes “(t)he number of customers accepting market offers”. 6 AGL, 2006 Full Year Financial Results, 16 August 2006, p. 17


2 8

• undertake a top-down review of operating costs by reference to the decisions by regulators and the available market evidence.

For the purposes of this exercise we focus on the bottom-up approach and cross check the results of this approach against regulatory decisions that have estimated retail operating costs.

EnergyAustralia is satisfied with Frontier’s overall methodology. We are, however, somewhat concerned with certain aspects in the delivery of the approach. In the remainder of this response, we will detail the refinements we believe Frontier should effect before finalising its report.

3.4.1 Cost build up

The Minister’s Terms of Reference require the operating cost allowance be based on a MMNE’s costs. In light of the Minister’s direction, EnergyAustralia engaged the help of KPMG to advise as to how a MMNE might be defined and some of the costs they might face. KPMG developed a detailed cost-build up model of a MMNE, drawing on its Benchmarking Study to ascertain reasonable discrete cost inputs that build to an appropriate cost allowance. KPMG’s analysis suggested a range of between $86 and $91 per customer per annum in 2006-07 dollars7. This range is considerably higher than that recommended by Frontier.

The key difference in its approaches is that KPMG relied on international data from its Benchmarking Study on utilities (consistent with the MMNE definition and average market parameters in operating in NSW) whereas Frontier relied on the standard retailers historical cost data.

In principle, EnergyAustralia does not disagree with the use of standard retailers cost data to provide some indication of an appropriate cost allowance. What we do contend is that this information should not be presented in isolation and then form the basis of the cost range recommendation. Frontier need to recognise and allow for the following:

•

•

•

EnergyAustralia’s operating costs are expected to increase from historical levels, particularly in light of labour cost increases in excess of inflation.

EnergyAustralia enjoys cost synergies with its distribution businesses that would not accrue to a MMNE.

As an incumbent, EnergyAustralia does not have costs associated with “getting into the NSW market”. These are more than simply customer acquisition costs.

7 Brought forward from $83 to $88 in 2005-06 dollars using Frontier’s assumed inflation rate of 3.2%.


2 9

Frontier should factor these differences into the cost recommendation. Failure to do so will result in a conflict with the Minister’s Terms of Reference.

Forward cost estimates

EnergyAustralia is surprised Frontier interpreted and discarded the standard retailers’ forward cost estimates in its analysis. In simple terms, its reasoning for ignoring the standard retailers’ future opex estimates is that both the fixed and variable costs components were observed to increase over time and Frontier’s expectation was that both these cost components would decline. As the standard retailers’ forecast did not match Frontier’s expectation, “limited weight” was given to these forecasts.

At no stage during Frontier’s review process were we asked to explain our forecast costs. The Information Request supplied did not seek this information either. In relation to operating cost data, the Information Request only sought explanation on how costs where allocated to “small customers”.

We believe it is worth exploring and understanding the issue of whether operating costs should be expected to change or remain constant over time. In the following sections, we will discuss the expectations of Frontier, EnergyAustralia, AGL / ESCOSA and Ofgem.

Frontier’s expectation

Frontier take the view that fixed operating costs should fall and variable operating costs should fall with a reduction in customer numbers. The support behind its assertion is quite vague and in some ways detached from the reality of mass market retailing. For example, Frontier argue that retailers can scale down customer information systems in response to changes in customer base size and enjoy fixed cost savings as customer numbers fall. But practically, a mass market retailer could not scale back a CIS in the short- to medium- term8 in order to achieve the theoretically saving Frontier envisage. ESCOSA recognised this point in its most recent review of AGL SA’s retail price path:

“as customers leave AGL SA for other retailers, the average cost per customer will rise given that some proportion of the operating costs is fixed (at least in the short to medium term)”

9. [Emphasis added]

Further on, Frontier argue on the one hand that short-term fixed costs actually become variable costs in the longer term, explaining why fixed costs should fall over time. On the other hand, Frontier claim that variable costs should fall over time, ostensibly ignoring those costs moving from fixed-to-variable on the overall level of variable costs.

8 That is, within the three year regulatory period. 9 ESCOSA, Inquiry into Retail Electricity Price Path Final Report, March 2005, p. 53.


3 0

In the final report, it will be important for Frontier to better evidence its expectation of forecast cost trends. This might be best achieved with some reference to actual results.

EnergyAustralia’s expectation

A large proportion of EnergyAustralia’s operating costs relate to labour10. Even costs like IT system maintenance are indirectly driven by labour input costs. EnergyAustralia expects labour costs will continue to rise in excess of CPI. This is supported by the forecast work performed by BIS Shrapnel which suggests that for the next three financial years, the Wage Cost Index is expected to increase by between 4.6% and 5.7% p.a.11

A significant proportion of our forecast costs relate to investing in our customer information system. The growing trend towards time of use pricing, and the associated system requirements, should not be underestimated12. Moreover, time of use pricing is relatively new to the mass market and the associated ‘teething’ problems (in terms of data handling, sharing, etcetera) should be considered. For this, and other wage-related reasons, we expect our customer information system costs to increase by more than 20% from actual costs during the 2004-07 Determination period.

Frontier note that it expects improvements in productivity, resulting in real decreases in overall operating costs. Although EnergyAustralia agrees that programs of efficiency (such as further outsourcing) are likely to place downward pressure on operating costs, we believe most of these gains have been exhausted since the commencement of Full Retail Contestability.

10 This is common among retailers. For example, ESCOSA noted that “a significant proportion of retail operating costs are attributable to salaries.” ESCOSA, Electricity Standing Contract Price, Price Determination, December 2004, p. A-11. 11 BIS Shrapnel, Economic Outlook, January 2007, p. 2. 12 Single consumption tariffs require 4 quarterly reads. Time of use tariffs, on the other hand, are calculated on 17,520 reads (ie. 48 half-hourly reads for 365 days of the year)


3 1

Ofgem’s expectation

In 1999, Ofgem determined an allowance for business supply costs per customer for each Public Electricity Supplier (PES)13. Three years later, in a paper on electricity supply competition, Ofgem noted the former PES costs actually increased by between 5 and 10% points14. Although real efficiency gains were made, these were more than offset by cost increases in areas such as bad debts, customer transfer and customer care.

In sum, we believe there is enough evidence to support real increases in operating costs over time and Frontier’s recommended operating cost allowance should reflect this.

Cost synergies

All three standard retailers are ‘stapled’ to a network business. Considerable corporate overhead costs, particularly in relation to customer information systems, are shared between the retail and network businesses. The MMNE, on the other hand, and as defined by Frontier, is a stand-alone retailer. The cost synergies available to a retailer / distributor are not available to a stand-alone MMNE retailer. To its credit, Frontier recognise this point: “Using the standard retailers’ reported costs as a proxy for the retail costs of a MMNE therefore risks understating the retail costs that a stand-alone MMNE would face.” Unfortunately, Frontier do not take the next logical step and attempt to estimate the quantum of this benefit, arguing “it is difficult”.

EnergyAustralia appreciates the difficulty associated with attempting to define the extent of the benefits available through economies of scope, especially in the absence of the necessary cost information. We are prepared to provide Frontier with any cost information they may need to make this assessment before finalising its report.

Market entry costs

As incumbents, the three standard retailers do not face the market entry costs that a MMNE would have to bear in entering the NSW market. These costs are not insignificant and include licensing, marketing (branding), and adhering to jurisdictional-based regulatory requirements such as the provision of certain information on bills, adhering to different metering protocols, etcetera.

13 Ofgem, Reviews of Public Electricity Suppliers 1998 to 2000 Supply Price Control Review. Final Proposals, December 1999, p. 30, Table 6.4. 14 Ofgem, Electricity supply competition: An Ofgem occasional paper, December 2002, p. 12.


3 2

Some of those market entry costs are associated with developing new business systems. For example, a NSW retailer must accept credit card payments by telephone (Electricity Supply (General) Regulation clause 30(2)(c)). A new entrant would need to establish systems to accommodate this payment channel.

A retailer entering the Victorian market must include on a customer’s bill “information concerning greenhouse gas emissions connected with the generation of electricity supplied to the customer or the generation of electricity in general, in accordance with an applicable guideline” (License condition 8.2(d)). NSW bills must report average daily consumption (Electricity Supply (General) Regulation clause 31(1)(e)). While the requirements are inconsistent, it is notable that NSW retailers participating in the Victorian market have made modifications to their billing systems to report both categories of information to all customers.

There are myriad other subtle but significant differences between state retail regulatory requirements – for example, the suite of “community languages” in which the retailer is required to provide interpretive services do not align between NSW and Victoria. Conducting investigations and amending business processes to ensure compliance with these regulations presents a significant cost to the new entrant retailer.

By using the standard retailers’ historical cost information as a proxy for a MMNE costs, Frontier effectively ignores these costs. EnergyAustralia submits that Frontier should attempt to identify and calculate ‘market entry’ costs and include these costs in its recommended allowance. Doing so will ensure that Frontier is consistent with the Minister’s Terms of Reference.

3.4.2 Benchmarking against regulatory audits

In its attempt to benchmark against regulatory audits, Frontier has used the results from:

•

•

•

Ofgem review of Public Electricity Suppliers (PES) in 1999;

CER review of PES in 2005; and

ESCOSA’s audit of AGL’s accounts in 2004.

It is important that Frontier uses the most relevant benchmarking data available in a manner that is appropriate. Failure to do so will undermine the credibility of the benchmarking exercise. With this in mind, EnergyAustralia would like to further explore the benchmark audits used by Frontier.

Ofgem review of PES

Frontier’s draft report includes a table of average operating costs per customer for the PESs in Ofgem’s jurisdiction. The table also includes a conversion to Australia


3 3

dollars per customer, taking into account inflation to bring the values to today’s dollars.

The Ofgem proposal report is quite dated, and EnergyAustralia believes that where there is more recent Ofgem benchmark data available to Frontier, this should be incorporated into its analysis.

Should Frontier continue to rely on this Ofgem report, the data available in the report needs to be correctly interpreted and we do not believe this is the case. Specifically, EnergyAustralia notes that the costs per customer are only for the variable costs faced by the PESs. Fixed costs are not included in these figures. A closer review of the report reveals that these costs are considerable and relate to the allocation transfer from the distribution business to the supply business. Consequently, the Ofgem average range estimate that Frontier relies on to establish “broad consistency” with its own recommended allowance is understated. Ofgem present an alternative table in the same report15 that includes a number of other adjustments. This table is presented below, converted to Australian 2006-07 dollars using Frontier’s conversion ratio. It indicates an average cost allowance of $114 per customer per annum is a more comparable benchmark value.

98/99, Unadjusted 98/99, Adjusted per customer

PES £ $ (06-07) £ $ (06-07)

Eastern 20.18 68 29.86 101

East Midlands 19.85 67 31.91 107

London 26.50 89 31.00 104

Manweb 20.12 68 34.05 115

Midlands 23.55 79 32.21 108

Northern 30.03 101 35.80 121

NORWEB 26.77 90 32.09 108

SEEBOARD 24.23 82 32.97 111

Southern 19.56 66 29.32 99

SWALEC 25.90 87 41.74 141

South Western 18.71 63 34.05 115

Yorkshire 29.35 99 32.67 110

Scottish Power 26.18 88 31.58 106

Hydro-Electric 51.99 175 45.04 152

Average 25.92 87 33.88 114

We believe Frontier’s analysis of Ofgem’s work should recognise the fixed charge component absent from the data presented in its draft report.

15 Ofgem, Reviews of Public Electricity Suppliers 1998 to 2000 Supply Price Control Review. Initial Proposals, October 1999, p. 75, Table 7.8.


3 4

CER review of PES

Frontier also included the benchmarking work performed by the Commission for Energy Regulation (CER) in Ireland for the PESs. The benchmarking is said to include survey participants from the US and Europe and resulted in a range of approximately €16 to €57 per customer in 2004.

Neither Frontier nor the CER proffer an explanation for the vast degree of range. Although presumably there are differences in operating efficiency levels between survey participants, these efficiencies alone could not account for the degree of difference between the two outlying companies. An alternative, which we put forward, is that there are enormous differences in the operating environment16 of the participants depending on their country of origin. Moreover, there would have been differences in accounting policies from country-to-country at the time the survey was conducted.

We argue that caution should be exercised when interpreting the results of the CER survey. We know little of the participants structure, size, extent of integration and other factors that will measurably impact on the operating cost of surveyed businesses.

ESCOSA’s audit of AGL’s accounts

As part of its price review, ESCOSA had the financial accounts of AGL audited. The audit captured actual costs for 2003, and was extended to include “2004-05 budgeted expenditure and forecast expenditure for 2005-06”.17

We feel that the ESCOSA review is the most relevant assessment presented by Frontier as it:

•

•

•

is based on an electricity retailer of known scale and scope operating in a domestic market;

was extracted directly from audited accounts; and

is the most recent assessment (when including budget and forecast costs) of the three benchmark audits.

Consequently, EnergyAustralia believes that it is this audit that should be given the greatest regard or ‘weighting‘ by Frontier when developing its recommended cost allowance.

16 This is a broad term and captures differences in regulatory and legislative obligations, average wages, license conditions, to name but a few. 17 ESCOSA, Inquiry into Retail Electricity Price Path: Final Report, March 2005, page 78.


3 5

3.4.3 Benchmarking against regulatory allowances

Frontier also presents recent regulatory decisions in Australia. These benchmarks provide the most relevant point of comparison for two main reasons: the decisions are recent (within the last three years) and are effected for businesses operating in the Australian retail electricity market. Frontier also appropriately present the cost benchmarks in today’s dollars.

EnergyAustralia notes that for some jurisdictions, Frontier have included more than one decision. EnergyAustralia fails to see the benefit in including benchmarks that are not the most current for each jurisdiction. There is clearly an upward trend in regulatory allowances – an issue addressed earlier in this report. But for the purposes of presenting a benchmark range, EnergyAustralia believes that only the most recent decision of each jurisdiction should be used.

When older, less relevant decisions are excluded from Frontier’s benchmarking graph and international benchmarks are included18, it becomes visibly evident that Frontier’s recommended range is considerably below allowances set in other jurisdictions.

18 Given the wide benchmark range for Ofgem (1999) and CER (2002) , only the middle third of the ranked results are shown. This is consistent with the approach taken by Frontier in its margin analysis, where the range of expected margins was quite wide.


3 6

EnergyAustralia believes that the inclusion of customer acquisition costs in the benchmarking results is misleading. The fact that regulators have not explicitly included these costs in their decisions, in EnergyAustralia’s view, quite clearly indicates that these costs were outside the scope of the regulators’ review. The Minister’s Terms of Reference takes a departure from other regulatory reviews and required that IPART recognise customer acquisition costs as costs specific to a MMNE entering the NSW market. It seems nonsensical to EnergyAustralia to argue the possibility that customer acquisition costs may have been included in other regulatory decisions when these costs have note been mentioned.

3.4.4 Benchmarking results compared to recommended allowance

EnergyAustralia submits that Frontier’s allowance for operating costs per customer is materially lower than the benchmarking results. Accordingly, Frontier’s recommended allowance should be adjusted upwards.

3 .5 MMNE re ta i l marg in

In its draft report, Frontier adopt the following approach to setting a recommended retail margin:

•

•

•

•

A bottom-up approach using Integral / NERA assumptions;

A bottom-up approach using standard retailers’ asset and revenue information;

An expected returns approach; and

Benchmarking regulatory decisions.

Frontier’s methodology is not dissimilar to the approach taken by KPMG in its review:

There are two main ways to benchmark the margin a MMNE might reasonably expect to earn:

• undertake a bottom-up cost build-up to determine the asset base and a benchmark return on that asset base to determine what that calculation implies for margins; and

• undertake a top-down review of margins by reference to the decisions by regulators and the available market evidence..

KPMG places greater emphasis on available market evidence and decisions by regulators, and largely dismisses the bottom-up review as an approach more suitable for an asset-intensive business and not an electricity retailer.

EnergyAustralia maintains the position in its submission that the appropriate level of retail margin can be determined with reference to available market data. The market evidence is likely to provide the best available indication of the sustainable margin a


3 7

MMNE might reasonably expect to earn, which might justify the decision to enter a market. Use of market data therefore focuses on real evidence that addresses the level of a sustainable margin to a MMNE Retailer.

EnergyAustralia is satisfied with Frontier’s overall methodology. We understand the ‘expected returns’ approach is unorthodox in a regulatory setting but appreciate that the results are benchmarked against other regulatory decisions for reasonableness.

EnergyAustralia believes the retail margin should compensate retailers for costs not compensated elsewhere in the framework. Our largest concern, as noted in our response on energy costs, is the absence of a number of residual risks from the recommended allowance of both energy costs and net margin. These risks include, but are not limited to, contract price movement, liquidity risk, wholesale (counter-party) credit risk, regulatory risk, demand risk, forecasting risk and hedge (contract) mismatch. We understand Frontier may have assumed these residual risks were to some extent captured in the standard retailers’ forecasts of pool prices and hedging instruments. This is not the case. EnergyAustralia is prepared to resubmit these forecasts in light of better understanding its application in Frontier’s modelling.

EnergyAustralia maintains that we are largely indifferent to where costs are allocated, as long as these costs are not missed. At this stage of the review we believe that the residual risks we have identified are best reflected in the energy cost allowance.

In the remaining sections, we provide some feedback we believe is of relevance to Frontier as it finalises its report.

3.5.1 Bottom-up approach

EnergyAustralia considers that a traditional bottom-up approach not suitable for measuring for retail margin, as would be conducted in an asset-intensive business. An electricity retail business does not have a large inventory of tangible assets on which to conduct a meaningful asset-based cost build up. KPMG recognises that this renders the bottom-up approach “problematic”19. As a result, the bottom-up approach is useful only to the extent that it is relied on as a ‘reasonableness check’.

Notwithstanding, we had asked KPMG to perform a cost build-up approach to better inform us of how such a calculation should be performed in the advent that IPART endorsed such an approach. The results of KPMG’s bottom-up analysis suggest that a net margin of between 4.3% to 7.1% is appropriate for a MMNE.

19 Much of the value of retail businesses would appear to lie in the more intangible assets of the business (ie. its customer base), rather than in the underlying physical assets.


3 8

In reviewing KPMG’s work, it is clear there are material differences between some of the assumptions / workings of KPMG and Frontier. For the purpose of providing feedback to Frontier on its draft report, we will concentrate the following discussion on these key points of difference.

Weighted Average Cost of Capital (WACC)

Frontier appear to rely on the WACC value used by Integral / NERA and benchmark this value against a limited number of reviews. KPMG, on the other hand, assumed a ‘ground-up’ approach to developing an appropriate WACC. A summary table of parameters from KPMG’s report is reproduced below:

For the sake of comparison, the pre-tax real WACC range that results from KPMG’s analysis is around 8.1% to 12.8%20 - significantly higher than the WACC value used by Frontier.

We believe that a pre-tax real WACC of 8% is on the low side for a retail business operating in a much ‘riskier’ environment than a typical, highly-regulated network business.

EnergyAustralia notes that net margin varies significantly with relatively small changes to the discount rate. Given its material effect, we submit that Frontier should place greater emphasis on determining the discount WACC with reference to its individual components such as cost of debt and level of gearing.

20 Updated for current average Commonwealth Government bond yield and implicit inflation.


3 9

Working capital

Managing working capital is an important element of the financial costs faced by our retail business. In calculating the net working capital of the retail business, it is necessary to direct close attention to the timing of receipts and payments.

By simply accepting NERA’s working capital assumption of one month, Frontier have not recognised impact of seasonal fluctuations in demand for electricity on a MMNE retailers’ working capital requirements. Electricity consumption is variable throughout a year and varies with the weather, customer lifestyle, economic influences and consumption patterns driven by installed appliances. In particular, consumption varies in a reasonably consistent way across a ‘typical’ year. Therefore, EnergyAustralia believes it is appropriate to include a ‘seasonality factor’ which increases the value of the required working capital to cater for seasonal fluctuations during the year. EnergyAustralia believes this seasonality factor is in the order of 35% above the flat average for each year. We are prepared to provide this information to Frontier for its consideration before finalising its report.

3.5.2 Expected returns approach

EnergyAustralia appreciates that the expected returns approach proffered by Frontier is firmly grounded in corporate finance theory. Adopting this approach takes a significant departure from the ordinary practice of other regulators in developing a regulatory retail margin allowance. Taking a well-established theory and applying it in a regulatory context brings inherent uncertainty. Consequently, we advocate that this (like any) untested method should always be supported by reference to other approaches to ensure sensibility. Frontier have achieved this in its multi-faceted approach.

3.5.3 Benchmarking approach

EnergyAustralia believes the role of benchmarking regulatory decisions should be limited to ensuring a ‘sensible’ outcome. The circularity of cross-referencing decisions made in other jurisdictions, combined with the ambiguity over which elements are in the margin calculation and which are not, limits the applicability of jurisdictional benchmarking to simply a ‘useful check’.

EnergyAustralia supports Frontier’s assertion that greater emphasis should be placed on the determinations of South Australia and Victoria. Here, the range of net margin (recommended or allowed) is between 5% and 8%. Other states have the benefit of mechanisms (such as vesting contracts) which limit the degree of exposure to volatile wholesale energy costs. Under the operation of ETEF, IPART provided a net margin allowance of 2% in its last retail Determination, recognising the limited risk standard retailers faced in their wholesale purchase costs for regulated customers. As EnergyAustralia and the other standard retailers are moving towards an environment


4 0

where ETEF will no longer be available, it is important that IPART recognise the additional burden of risk.


4 1

4 A P P E N D I X A – K P M G R E P O R T F O R E N E R G Y A U S T R A L I A O N B E N C H M A R K I N G R E T A I L O P E R A T I N G C O S T S A N D M A R G I N S

ABCD

EnergyAustralia

Benchmarking Retail Operating Costs and Margins

September 2006 This report contains 57 pages

EA06-RetBmMkRH1015-SAR1.doc

© 2006 KPMG, an Australian partnership, is part of the KPMG International network. KPMG International is a Swiss cooperative. All rights reserved. The KPMG logo and name are trademarks of KPMG.

ABCD EnergyAustralia

Benchmarking Retail Operating Costs and MarginsSeptember 2006

ii © 2006 KPMG, an Australian partnership, is part of the KPMG International network. KPMG International

is a Swiss cooperative. All rights reserved. The KPMG logo and name are trademarks of KPMG.

Contents

1 Introduction 1 1.1 Outline of approach 1 1.2 Outline of report 1 1.3 Disclaimer 2

2 Approach 3 2.1 IPART’s Issues Paper 3 2.2 Definitions 4

3 Retail operating costs 8 3.1 Our approach 8 3.2 Electricity retailing 8 3.3 Defining the operating costs of a mass market new entrant 9 3.4 Results 10

4 Retail margins 11 4.1 Recent regulated electricity retail tariff decisions 11 4.2 Market evidence 13 4.2.1 Independent Experts’ reports 14 4.2.2 Company and brokers’ reports 15 4.2.3 International evidence 16

5 Findings and commentary 18 5.1 Qualifications 18 5.2 Potential NSW market issues 19 5.3 Other issues in setting appropriate regulated retail tariffs 21

A EnergyAustralia’s terms of reference 25

B Operating cost benchmarking 26 B.1 Business assumptions 26 B.2 Cost tree 28 B.3 Data, benchmarks and calculations 34 B.4 Summary 51 B.5 Comparative analysis 51

C Recent regulated electricity retail tariff decisions 53



1© 2006 KPMG, an Australian partnership, is part of the KPMG International network. KPMG International


1 Introduction The Independent Pricing and Regulatory Tribunal (“IPART”) is currently engaged in a process to determine regulated electricity retail tariffs for the period 2007 to 2010. As part of this process, IPART recently published an Issues Paper which calls for submissions from interested parties on the issues involved in setting the tariffs.1

EnergyAustralia (“EA”) has engaged KPMG to benchmark the:

• operating costs of a mass market new entrant (or “MMNE”); and

• retail margins a MMNE might reasonably expect to earn.

1.1 Outline of approach To undertake this assignment we have:

• reviewed IPART’s Issues Paper;

• defined the entity for which we are benchmarking operating costs and margins;

• benchmarked operating costs and margins by reference to a range of evidence; and

• examined the findings and commented on how to interpret them.

1.2 Outline of report This report provides the output of our analysis. In particular:

• section 2 outlines how we have approached the task;

• section 3 outlines how we have benchmarked retail operating costs;

• section 4 outlines how we have benchmarked retail margins; and

• section 5 provides our findings and comments.

There are three appendices:

• appendix A summarises our Terms of Reference;

• appendix B describes how we have benchmarked retail operating costs; and

• appendix C describes other regulatory decisions in respect of operating costs and margins.

1 IPART, Review of Regulated Retail Tariffs and Charges for Electricity 2007 to 2010, July 2006.





1.3 Disclaimer

Inherent Limitations

This report has been prepared as outlined in section 2. The procedures outlined in this report constitute neither an audit nor a comprehensive review of operations.

No warranty of completeness, accuracy or reliability is given in relation to the statements and representations made by, and the information and documentation provided by, EnergyAustralia consulted as part of the process.

KPMG have indicated within this report the sources of the information provided. We have not sought to independently verify those sources unless otherwise noted within the report.

In the course of our work, projections have been prepared on the basis of assumptions and methodology which have been described in our report. It is possible that some of the assumptions underlying our projections may not materialise. Nevertheless, we have applied our professional judgement in making these assumptions, such that they constitute an understandable basis for estimates and projections. Beyond this, to the extent that certain assumptions do not materialise, then it must be appreciated that our estimates and projections of achievable results will vary.

KPMG is under no obligation in any circumstance to update this report, in either oral or written form, for events occurring after the report has been issued in final form.

The findings in this report have been formed on the above basis.

Third Party Reliance

This report is solely for the purpose set out in section 1 of this report and for EnergyAustralia’s information which includes the use of this information in EnergyAustralia’s response to the IPART Issues Paper. It is not however, to be used for any other purpose or distributed to any other party without KPMG’s prior written consent.

This report has been prepared at the request of EnergyAustralia in accordance with the terms of KPMG’s engagement letter/contract dated 26 July 2006. Other than our responsibility to EnergyAustralia, neither KPMG nor any member or employee of KPMG undertakes responsibility arising in any way from reliance placed by a third party on this report. Any reliance placed is that party’s sole responsibility.





2 Approach This section outlines our approach to benchmarking the retail operating costs and margins of a MMNE. In particular, it summarises the relevant parts of IPART’s Issues Paper and describes how we have defined the relevant terms and the task.

2.1 IPART’s Issues Paper The Issues Paper raises and seeks comment on a number of matters relevant to the regulation of retail electricity tariffs. These include:

• policy changes impacting on the review;

• the form of regulation;

• costs to be recovered by regulated tariffs; and

• miscellaneous charges.

IPART considers that the ‘costs’ regulated tariffs ought to recover include the following:

• retail margin;

• retail operating costs; and

• costs of supply including:

- energy;

- network charges;

- green energy;

- hedging , risk management and transaction costs; and

- costs of new schemes that may be imposed.

In identifying the range of costs that regulated tariffs ought to recover, IPART seeks comment from interested parties on retail operating costs and margins (the two areas this report addresses). More specifically, IPART has sought comment on the appropriate:

…level of mass market new entrant retail operating costs for inclusion in regulated tariffs.2

…mass market new entrant retail margin to be included in regulated retail tariffs.3

2 Ibid., page 20. 3 Ibid., page 21.





2.2 Definitions IPART is required by the NSW Minister of Energy’s Terms of Reference to set regulated tariffs, in part, according to the operating costs and margins of a mass market new entrant (or “MMNE”).

A mass market new entrant

The Minister of Energy’s Terms of Reference does not define what it means by a “mass market new entrant” other than to state that a “new market entrant that is of sufficient size to achieve economies of scale”. Moreover, IPART does not explain how it interprets the term either, although at one point it uses the term “competitive” mass market retailer.4

This raises a number of questions that may be relevant to benchmarking operating costs and margins for a MMNE including:

• the market definition;

• the size and type of retailer;

• whether a “new entrant” implies extra costs (eg. associated with getting into the market);

• whether a “new entrant” implies a retailer that is not vertically integrated; and

• whether the task is to determine the margins a MMNE would expect to earn in a market that did not have price regulation.

These questions are important because they go to interpreting both the objectives of the Terms of Reference and the evidence benchmarking produces, which inevitably requires considerable judgment.

The objectives of the Terms of Reference appear to require new entrants to be able to compete on a ‘level playing field’ with regulated tariffs. The Terms of Reference states that IPART:

…must consider the Government’s policy aim of reducing customers’ reliance of regulated prices and the effect of its determination on competition in the retail electricity market.5

It also states that the determination should ensure that all retail charges are at cost reflective levels for all small retail customers by 30 June 2010.

In our view, a MMNE would have two key features. It would:

• retail electricity to a large customer base and to a broad range of customers; and

4 Ibid., page 21. 5 Ibid., page 30.





• would employ best practice operations, systems and financial structures that would enable it to compete effectively with reasonably set regulated tariffs and earn returns that justify the risks associated with entering into the market.

Table 2.1 outlines our definition of a MMNE.

Table 2.1: Defining a Mass Market New Entrant (MMNE) Definition Comment

Area of operation Assumed to be NSW for customers using less than 160 MWh per year, reflecting IPART’s jurisdiction to regulate prices.

Electricity retailer only Reflecting IPART’s price determination of electricity tariffs.

Stand alone business Assumed to be a stand alone, non-vertically integrated retailer. This will require the retailer to purchase all energy sold either via hedging contracts or via the spot market.

Customer numbers A retailer would need to maintain a customer base of at least 250,000 customers to warrant the investment in systems necessary to handle a mass market. We assume (for simplicity) that the MMNE is able to achieve this scale of operation immediately on commencement of operations (see below).

Customer types To operate in a mass market, the retailer should be in a position to accept all customer types including: - all metering types; - a range of customer types including small commercial and residential (including

those from a relatively broad socioeconomic range); and - all payment types (direct debit, cheque and cash, and those requiring reminder

notices and with payment difficulties). Call centre contacts When establishing costs for operating the call centre, we consider that it will be driven by

the number of customers, the complexity of tariffs and billing information, and externalities such as market information, government policy etc. We have therefore assumed that the MMNE will operate in a similar manner to that currently experienced in the market place.

Billing intervals Billing intervals should accommodate mostly quarterly read and bill cycles and a portion of monthly read and bill.

Handling of data Data to be processed should include customer data for customers with time of use metering and customers with accumulation metering types.

Section 5 discusses in some detail the implications of our definition in commenting on our findings. In most cases, these implications go to how to interpret the benchmarks. This is primarily because although the available benchmarks are typically consistent with what we are seeking to quantify, they often lack the precision necessary to enable use without the exercise of judgement.

A couple of comments are worth reiterating here.

The benchmark entity described as a “mass market new entrant” is in some ways a hypothetical construct, involving a number of simplifying assumptions, which are potentially significant. For example, we have assumed that the MMNE has 250,000 customers and has attained these customers immediately. In practice, the new entrant is unlikely to enter, in the first instance, as a mass player and it is likely to take a number of years to develop such a customer base. If it is to compete in the mass market it will, however, need to have the systems in place to do so and the capability of dealing with the complexities of having a diverse customer base





(notwithstanding that in the short term at least its customer base is likely to be less diverse). It will have the operating costs to match.

In benchmarking the operating costs of a MMNE we have excluded customer acquisition costs (which are of more of a capital cost in nature) but not the cost of having such a capability. Acquisition costs are relevant to the analysis, but we account for them in the appropriate net margin, which our benchmarking of net margins implicitly captures. We have also excluded incumbent retailer costs, because a MMNE would not bear these costs.

So our assumptions, while somewhat simplistic, are nevertheless consistent with the underlying assumption regarding a MMNE that is able to exploit economies of scale.

Retail margins

IPART’s Issues Paper defines the retail margin (or profit margin) as the reward to investors for committing capital to a business. This implies that it is the difference between the tariffs charged to a group of customers, and the costs incurred in selling energy to those customers.

IPART identifies a range of factors that will influence the retail margin including the level of risk associated with:

• energy purchasing costs;

• customer default and ‘typical’ bad debt expense; and

• competition from electricity substitutes (eg. gas, solid fuels etc).

IPART also identifies hedging, risk management and transaction costs and the forecasting risk that retailers will face in the absence of ETEF as potentially relevant to the estimation of the retail margin.

In principle, the margin should compensate a MMNE for the market risks associated with the underlying assets, and all the business risks should be compensated for in the operating cost allowances made for such a business. In practice, however, delineating risks and the compensation for them on this conceptual basis is typically problematic. This means that, in practice, it is always likely to be difficult to separate entirely the allowance for risks incorporated into:

• the operating cost allowances generally;

• the allowance for the costs of managing certain ‘business’ risks (eg. hedging costs)6; or

• the allowance for the market risk associated with the underlying assets.

6 Although we have included energy trading transaction costs in our operating benchmarks, it would not be unreasonable to account for these costs within the energy cost allowance as an alternative.





For the purposes of this exercise we have established operating cost benchmarks consistent with the available market information. Therefore:

• IPART will account for the costs of acquiring electricity elsewhere, but that the transaction costs associated with doing so are reflected in our benchmarks;

• our operating cost estimates reflect those ‘business’ risks that are readily quantifiable (this includes an allowance for ‘typical’ bad debts); and

• market risks (and presumably any other risks not captured above) are reflected in the margins it would be reasonable to expect a MMNE to earn.

For the purposes of this report we have used the following definitions of margins:

• margin: generally refers to the difference between the revenue and costs of a business. It is possible to express margins at a number of levels of a business’s Income Statement (see below).

• gross margin: is the difference between revenues and the direct costs7 of production of a business. In this case, we assume it means revenues less all electricity (commodity) costs and other pass through costs (eg. network charges).

• net margin: is the difference between the gross margin and the indirect costs8 of production of a business. In this case, we assume it means all operating costs, excluding interest and taxes.9

In addition, this report also uses EBITDA (“Earnings before interest tax depreciation and amortisation”) and EBIT (“Earnings before interest and tax”) margins because the market often expresses margins in these ways. The latter is broadly consistent with our definition of net margin. In practice, because of the nature of the retail electricity business, the differences between EBITDA and EBIT margins are typically relatively modest.

Unless otherwise indicated, this report expresses all these margins as a percentage of revenue.

7 Direct costs: generally refers to costs that are caused by the production of a particular product or service and may therefore be directly attributed to, or readily identified with, a particular product, service or contract (for example, raw materials used to produce a good). 8 Indirect costs: generally refers to the costs of doing business that are not directly related to a particular product, service or contract. It should be noted that determining what direct or indirect costs are in particular circumstances or for particular purposes can require a considerable degree of judgment, and estimating some margins is therefore subject to the same uncertainties. 9 As the net margin estimates the return on capital it implicitly includes the costs associated with the return on capital employed to acquire customers.





3 Retail operating costs There are two main ways to benchmark the operating costs a MMNE could expect to incur:

• undertake a bottom-up, cost build-up exercise; and

• undertake a top-down review of operating costs by reference to the decisions by regulators and the available market evidence.

For the purposes of this exercise we focus on the bottom-up approach and cross check the results of this approach against regulatory decisions that have estimated retail operating costs.10

3.1 Our approach This section summarises the bottom-up approach, which we describe in further detail in Appendix B. Our approach involves:

• defining the electricity retailing function and the scope of a MMNE’s business;

• identifying the activities that are likely to be involved in undertaking the function;

• identifying the types of resources involved in undertaking those activities; and

• identifying the quantum of resources and unit costs that are likely to be consumed.

To do this, we identify and rely on the best available market information and benchmarks in respect of both the quantum and unit cost of the required resources.

3.2 Electricity retailing Electricity retailers provide consumers with ways to buy electricity. From the customer’s perspective the electricity retailers’ offer has three key components:

• a price charged for access to electricity and for the amount of electricity consumed;

• a method by which customers pay; and

• typically, certainty in regard to the prices customers pay for a certain period.

Electricity retailers seek to add value by providing these services at a price and in a manner that meets customers’ needs. It involves the following key functions:

• sales and marketing (eg. to generate revenues by winning and retaining customers);

10 There is relatively little detailed information on operating costs per customer available from the capital markets.





• billing and revenue collection (eg. to generate cash inflow);11

• customer service (eg. to manage customer communication); and

• risk management (eg. minimise energy costs by managing purchase risk).

The first function is a profit centre, while the last three functions are principally cost centres.12

The retailers’ costs are primarily either fixed or customer-related. A retailer’s fixed costs are primarily a function of the information technology systems required. A retailer’s customer related costs are primarily a function of processing the needs of each customer (i.e. having the information available to bill the customer, sending bills, collecting revenues etc).

Electricity retailing to a mass market is a high volume business that involves significant fixed costs (eg. a CIS system for a mass market retailer is likely to cost approximately $15-20 million).

3.3 Defining the operating costs of a mass market new entrant Due to the significance of systems needed to establish and maintain a MMNE retailer operation and the nature of the associated costs, we have formed the view that a MMNE would need to be serving at least 250,000 customers to justify the necessary investment. Appendix B.1.3 explains our reasoning.

A cost tree in Appendix B.2 sets out the key activities of a MMNE. We have then determined the drivers of costs for those activities and how we expect a MMNE retailer to operate based on benchmarks and observations which appendix B discusses.

Table 3.1 below defines the activities, comments on the key cost drivers and identifies the benchmarks we use.

11 Sometimes the metering and or meter reading function are also included. 12 Energy trading can also be a profit centre, but that is a function that can and usually is undertaken separately to energy retailing (ie. it is similar to the distinction between stock broking and proprietary trading).





Table 3.1: Summary of cost drivers for the benchmarks used in the operating costs of a MMNE Activity Comments on cost drivers Benchmark

Billing, data validation

Driven by number of bills issued, which in turn is a function of the billing cycle (monthly or quarterly), plus reminder notices

Cost per bill

Collection Driven by number of customers and billing frequency Cost per receipt

Customer transfer Driven by the number of customers and transfer rates Cost per transfer

Bad debt expense Driven by size of business Electricity market experience based on revenue

Customer Information System

Driven by the size of the business. We have assumed a 250,000 customer base

Benchmark costs based on IT Utilities surveys

Call centre costs Driven by the number of customers and the number of calls customers make to their retailer

Benchmark costs built up from call centre statistics

Management

Pricing

Risk management

Settlements

Regulatory

Driven by size/scale of the business. We have assumed a size of business for a MMNE that requires the management infrastructure defined in this benchmarking exercise

Benchmark staffing costs and occupancy costs, plus minor amounts for licensing, and ombudsman.

Energy trading Driven by size of business Salary and occupancy costs

Public relations Driven by size of business, with an allowance for mass marketing budget

Salary and occupancy costs plus an advertising budget based on regulatory precedents

3.4 Results Application of the benchmarks against the MMNE assumed in this report produce the operating costs Table 3.2 summarises.

Table 3.2: Summary of results - operating costs of a MMNE as defined above

2005/06 Benchmark

Section in appendix

Lower bound benchmark

($m)

Upper bound benchmark

($m) %

Billing and customer collection (including CIS/ITS) B.3.1 10.04 10.19 46-50 Call centre costs B.3.2 2.05 2.05 9-10 Full time equivalent employee costs and overheads B.3.3 8.57 9.77 41-44

Total 20.70 22.00 100%

The total retail operating cost for the MMNE ranges from $20.7m to $22.0m, resulting in a per customer cost of about $83 to $88 for 2005/06.





4 Retail margins There are two main ways to benchmark the margin a MMNE might reasonably expect to earn:

• undertake a bottom-up cost build-up to determine the asset base and a benchmark return on that asset base to determine what that calculation implies for margins; and

• undertake a top-down review of margins by reference to the decisions by regulators and the available market evidence.

For the purposes of this exercise, and for the reasons we illustrate, we use the latter approach.

4.1 Recent regulated electricity retail tariff decisions The decisions by regulators might provide some evidence of what a MMNE might expect to earn.

We have summarised below the results of regulatory decisions on retail margins. Appendix C provides further details on these decisions.

Table 4.1: Jurisdictional decisions - margins

Jurisdictional Benchmarking

Year Net margin %

SA Electricity 13 2005 5% NSW Electricity 2004 2% Victoria Electricity 2003 5% SA Electricity 2003 5% Tasmania Electricity 2003 3% ACT Electricity 2003 5% NSW Electricity 2002 1.5 to 2.5% Victoria Electricity 2001 2.5 to 5% NSW Electricity 2000 1.5 to 2.5% Tasmania Electricity 1999 1.5%

It is perhaps unsurprising that there has been more competitive activity in those markets where regulated margins are higher. For example, there has been a considerable degree of competitive activity and some new entry into the Victorian and South Australian markets. According to AGL, annual customer churn rates for Victorian and South Australian gas and electricity are all at least 20%, whereas for NSW gas and electricity the churn rates are 4% and 9% respectively.14 In addition, new entry into the NSW market has been modest.

13 See page 53. 14 AGL, 2006 Full Year Financial Results’, 16 August 2006, page 17. The SA figures include retentions whereas the others do not and the Government also subsidised customer switching.





This suggests that the regulated retail prices in Victoria and SA have been sufficient to encourage switching and new entry. This does not necessarily mean, however, that the allowed margins are the primary reasons for this. For example, lower than expected wholesale electricity prices would appear to be partly responsible for some of this competitive activity.15 Our discussions with market participants have confirmed this view.

This view would appear to accord with the market evidence.

In the capital markets one of the most commonly used measures of the relative cost of acquiring retail businesses is the cost per customer.16,17 Most major transactions, many of which occurred some years ago, appear to have valued customers in the $400 to $850 range (with the higher end mainly for customers for whom dual fuel offers are an option). More recently, somewhat higher valuations would appear to be the norm.

For example:

• ABM Amro values AGL’s retail business at approximately $925/customer;18

• Morgan Stanley values Origin’s retail gas business at $900 per customer and its retail electricity business at $980 per customer.19

Goldman Sachs JBWere also recently stated that it expected ENERGEX’s retail business to sell for about $600 million, which works out to be approximately $750 per electricity customer.20

To justify these valuations the market would appear to be expecting these customers to generate margins above those indicated by most regulated margins.

For example, net margins of 3% on an electricity bill of $1000 per annum imply that the retailer is earning (before interest and tax) about $30. The above valuations would suggest that at $900 per customer the market is prepared to pay about 30 times these earnings (ie. it would take the

15 For example, Victorian electricity prices in the 2004/05-2005/06 appear to have been lower than expected around the time the regulated price path was set. The Age, “Transfield gets Loy Yang smoking’ 16 August 2006. It states “Transfield Services' purchase of a 9.3 per cent stake in Victoria's biggest generator, Loy Yang A, for $115 million earlier this month shows how generation asset values have risen in recent times.” And “The reason the price has crept up is that for the first time in a while, power prices are actually tracking in the right direction as far as generators are concerned. In 2003-04, average Victorian power prices bottomed at $25.38 per megawatt-hour. A year on they climbed to $27.62, then last year to $32.47. For the first month of the current financial year they have averaged $46.39. By contrast, in October 2003 the expectations were for prices of around $33 and $35 in 2004-05 and 2005-06 respectively. See ESAA, Market Report, Week 40, October 2003. This also appears to have encouraged the Government to seek further reductions in retail prices over the regulated period. See Power Industry News, ‘Vic Price Boast’, Edition 494, 5 June 2006, page 18. 16 Australian Financial Review, ‘Australian Energy poised’, 6 August 2003, page 27. 17 Australian Financial Review, ‘AGL needs up to $500m equity for Pulse’, 1 July 2002, page 16. See also the Allen Consulting Group, Review of the Gas Code: Commentary on Economic Issues, Report to BHP Billiton, August 2003. The paper provides a summary of Australian retail energy business transactions. 18 ABN Amro, Australian Gas Light’ 28 February 2006, page 5. 19 Morgan Stanley, ‘Origin Energy Limited’, 21 February 2006, page 5. 20 Australian Financial Review, Alinta-AGL deal to power activity, 23 August 2006, page 33. Although it also has about 50,00 LPG gas customers as well.





same number of years to recover the cost of paying this much to acquire a customer).21 At a net margin of 5%, the market is prepared to pay 18 times these earnings.

It seems unlikely that this is the case. For instance, ABN Amro suggests that it values AGL’s retail business on an expectation that it earns an EBITDA of approximately $100 per customer, and that the market is willing to pay 9 times EBITDA to acquire this performance. Moreover, AGL’s most recent results suggest it is earning a gross margin of $184 per electricity mass market customer.22 As section 4.2 illustrates, these figures are consistent with net margins above those implicit in regulated tariffs.

The evidence suggests that the margins within regulated retail electricity tariffs are likely to be at the low end of those necessary to encourage new entry, particularly for a MMNE.

4.2 Market evidence There are a number of sources of market evidence on expected retail margins including:

• independent experts’ reports; and

• brokers’ reports (which often refer to company results).

Similar sources of information also available in relation to international data.

The key advantage of this evidence is that it provides the most direct indication of what the market expects from the performance of energy retailers. The key disadvantage of this type of evidence is that it often relates to the market’s expectations of the performance of a particular business, or of the market in particular circumstances. In other words, often the information relates to retailers that are:

• part of a larger diversified utility business;

• undertaking a number of activities in addition to energy retailing (sometimes within the business unit containing energy retailing);

• much larger than a MMNE could reasonably expect to be (who are often large players in the lower margin, higher volume commercial and industrial markets as well);

• diversified across a number of regions (and/or countries) and fuels (eg. including LPG) within the energy retail business; and

• vertically integrated in the electricity (and possibly gas) sector.

21 This evidence also explains why estimating margins according to bottom-up approach can be problematic. Much of the value of retail businesses would appear to lie in the more intangible assets of the business (ie. its customer base), rather than in the underlying physical assets. 22 AGL, 2006 Full Year Financial Results: 12 months ended 30 June 2006, page 15. It reports a gross mass margin of $263.4 million over 1.428 million accounts.





These caveats by definition draw into question whether the information provided from these sources is relevant for a MMNE. For example, a MMNE might require a higher margin than the incumbents to justify the risk of entry.

In addition, the information provided by brokers is often incomplete or may not be consistent with the benchmarks we estimate.23

As a result, it does not obviate the need to exercise judgement when interpreting the evidence. That said, we are of the view that the market evidence is likely to provide the best available indication of the sustainable margin a MMNE might reasonably expect to earn, which might justify the decision to enter a market. We have focussed on evidence that goes to sustainability.

On balance, the evidence set out in the following sections suggests that a MMNE into a competitive retail electricity market might reasonably expect net margins in the order of 5-8%.

4.2.1 Independent Experts’ reports Independent experts’ reports are usually prepared in the context of demergers and certain takeovers, where a Scheme of Arrangement is required. There have been two recent independent experts reports produced that are worth noting:

• Grant Samuel’s report in respect of the (original) AGL demerger proposal; and

• BDO Kendalls’s report in respect of the takeover of Australian Energy Ltd.

The AGL demerger

In October 2005 AGL proposed a demerger of its infrastructure and energy businesses (which has since been replaced by another transaction between it and Alinta, which achieves a similar outcome). Facilitating a demerger requires a Scheme of Arrangement. The Scheme requires an independent expert’s report confirming that it is in shareholders’ interests. The AGL Board retained Grant Samuel for this purpose.

Grant Samuel’s report notes that:

• AGL Energy (would have been) a large listed Australian energy company with operations predominantly in eastern Australian encompassing:

- energy generation and production assets; and

- the marketing and retailing of energy.

• AGL Energy will continue its full service (ie. vertically integrated) energy model; and

• AGL Energy will be Australia’s largest energy retailer with 2.8 million energy customer accounts (of which more than one million are dual fuel accounts).24

23 For example, it is not always obvious whether the margins contain an allocation of corporate overheads or not in instances where the retailer is part of a larger business. 24 After the new merger with Alinta’s retail business, the figure is likely to be closer to 3.5 million customers.





Its estimates of AGL Energy’s Pro Forma Financial Performance suggested margins for the year ended 30 June 2005 of:

• 8.5% EBITDA margin; and

• 6.2% EBIT margin.

This covers both aspects of AGL Energy’s business so is a less relevant indicator of retail margins (although it is similar to its retail business results).

Australian Energy Ltd

In December 2005, Australian Energy Limited, trading as Powerdirect, announced that Ergon Energy proposed to acquire all outstanding shares and options in the company, via a Scheme of Arrangement.

Powerdirect focuses on selling electricity to small and medium sized businesses in the NEM (eg. petrol stations, retail outlets, residential apartments, commercial office buildings and water authorities). It has over 40,000 customers in Victoria, representing 73.5% of total volume sold. It is not vertically integrated and is not a mass market retailer.

Its consolidated statements of financial performance for the 12 months to June 2004 and June 2005 respectively, indicate the following margins. Table 4.2: Australian Energy Ltd - margins25 30 June 2004 30 June 2005

Gross margin 15.6% 13.9% EBITDA margin 5.4% 4.5% EBIT margin 5.1% 4.3% Net margin (after interest and tax) 4.9% 4.1%

4.2.2 Company and brokers’ reports We have reviewed company reports and a sample of the recent brokers’ reports available to us.26 The information focuses on the two listed major retailers in Australia (ie. AGL and Origin).

On the 16 August 2006 AGL announced its full year results. It included for the retail business an increase in EBIT to sales margin to 7.7% from 6.6% (and a 2006 EBITDA of 8%).27 The result was based on gross margins of 13.2% for its electricity business.

An earlier report by ABN Amro expected an EBIT margin of 7% for the full year for AGL, broadly in line with the previous calendar period.28 It also argues that “retail margins are showing signs of being sustainable”.

25 KPMG has estimated the EBITDA and EBIT margins from the available data. 26 KPMG has access to two on-line sources of broker’s reports: Thomson and Onesource. 27 AGL, 2006 Full Year Financial Results, 16 August 2006, page 4. It is not obvious whether this is before or after an allocation of corporate overheads.





Origin Energy

Origin’s final results for 2006 suggest that it earned an EBIT margin in its Australian natural gas and electricity retail business (ie. excluding LPG) of 7.8% compared with 7.2% in 2005. This is $110 per customer.29 The LPG business made a lower margin (ie. about 4.5%). The gross margin for the electricity business was 17.3%.

CitiGroup states in relation to Origin Energy that its medium term expectations for the retail business remain for a stabilised EBIT margin around the 6% level.30

Morgan Stanley values Origin Energy on a future EBITDA/sales margin for its retail business in the 8.2-8.4% range.31 This suggests an EBIT margin of 6.8-7.0%.32

Meanwhile, JP Morgan estimates gross margins for Origin Energy’s retail electricity business in 2006 and 2007 of 16.3% and 16.1% respectively.33

4.2.3 International evidence We have also reviewed some of the available information in respect of the United Kingdom and New Zealand. A key advantage of this information is that these are two national retail electricity markets that have fully deregulated retail prices. They therefore provide evidence on the margins that retailers would expect to earn in such a market. A key disadvantage is that there are likely to be several factors influencing the margins retailers might expect to earn in these countries.

For example, the UK market displays a high degree of vertical and horizontal (ie. dual fuel) integration, while the NZ is highly vertically integrated. These features may reduce the required margins for the relevant retailers (see section 5.2 below). The UK is also typically assumed to have a lower cost of capital. New Zealand has a generally less volatile electricity market due to it dependence on hydroelectric power, but one that nevertheless produces occasional but severe imbalances in periods of extreme weather (ie. drought).

Nevertheless, this evidence provides another set of comparables.

New Zealand

Contact Energy is one of New Zealand’s major retailers (one of two that competes across the country) and is highly vertically integrated. Its retail performance in recent times has been highly volatile, but generally compensated for by its generation business. For example, in the year to June 2005, Contact retail business made an EBITF (before financial instruments as well,

28 ABN Amro, Australian Gas Light, ‘Solid result, but what happens next’, 28 February 2006, page 2. 29 Origin Energy, Directors’ Review of Results for the full year ended 30 June 2006, 30 August 2006, page 17. These figures appear to be after an allocation of corporate costs. 30 CitiGroup, Origin Energy Limited, ‘No Contact’, 28 June 2006, page 12. This includes the LPG business. 31 Morgan Stanley, Origin Energy Ltd., ‘2006 Interim Result and Merger with Contact’, 21 February 2006, page 5. 32 Origin’s interim results for 2006 resulted in EBITDA and EBIT margins from the retail segment falling from 9.8% to 8.8% and 8.4% to 7.4% respectively. 33 JP Morgan, Origin Energy, ‘Upstream costs moderate growth outlook’, 29 August 2005, page 3.





but after an allocation of corporate costs) of 14.9%. In the year to June 2006, however, the same margin was -11.1%. However, generation business margins increased substantially.34

CitiGroup assumes that retail EBIT margins for Contact Energy in New Zealand would fall to an average of 7.5% over the next few years.35 On CitiGroup’s analysis, this would be consistent with a gross margin of about 13.6%.

United Kingdom

There has been considerable volatility in the UK market in recent times due to upward pressure on wholesale electricity prices, which has lead to significant upward pressure on retail prices.

The company that has been most exposed to this uncertainty is Centrica. It has approximately 16.5 million domestic energy customers, but has relatively little vertical integration particularly in electricity (see below). Its retail energy division has recently been making significant losses.

In March 2006, Moody’s changed its outlook in respect of Centrica to negative despite its recent announcement to increase its tariffs by 22%. The change in outlook highlighted the risk of Centrica losing a lot of customers because of weaknesses in its hedging arrangements and thus its retail franchise. Moody’s stated that its reassessment:

…underlines Centrica’s current disadvantage over its UK peers from its short generation position and its gas-biased and thus costly fuel mix. As a result, Centrica’s residential energy margins are further impacted by the necessity to purchase, at current high prices, power to meet its requirements. Moody’s assumes that further acquisitions of power generating capacity (including power purchase agreements) are inevitable, but that such will come at a high price in the present environment, similarly to any upstream gas assets.36

We understand that by June this year Centrica had lost over 400,000 customers in the first half.

Recent brokers’ reports have suggested as follows:

We assume a long-term average (LTA) EBIT margin of 5.7% for this division;37

The key value drivers in our Base Case valuation are (1) Sustainable retail margins of 6%.38

Our revised valuation for Centrica is set out below. We are now assuming that stickier customers will mean 8% margins could be sustainable in the energy retail business.39

34 Contact Energy Limited, Management Discussion of Audited Consolidated Financial Results for the 12 months Ended 30 June 2006, pages 4-6. 35 CitiGroup, Origin Energy Limited, ‘No Contact’, 28 June 2006, page 4. 36 Moody’s Investor Services, ‘Moody’s Changes Outlook on Centrica’s Ratings to Negative’, 16 March 2006. 37 ABM Amro, ‘British Gas Services – upping the ante’, 27 June 2006, page 2. 38 Morgan Stanley, ‘Centrica – Recovery Fully Priced’, 27 July 2006, page 3. This appears to relate purely to the UK retail energy business. It noted, however, that Centrica has never achieved these margins. 39 Deutsche Bank, ‘Centrica’, 23 February 2006, page 4. This appears to be a Profit After Tax margin but appears to include the benefit of some generation assets within that business.





5 Findings and commentary Our key findings are as follows:

• Our benchmarking in section 3 suggests that a reasonable retail operating costs for a MMNE are likely to be in the range of $83-88 per customer; and

• Our benchmarking in section 4 suggests that an appropriate retail net margin for a MMNE is likely to be in the range of 5-8%.

Moreover, a margin at the upper end of this range is likely to be more consistent with meeting the Minister of Energy’s primary objective of having cost reflective tariffs and encouraging a MMNE, particularly in light of what we believe is the most appropriate interpretation of the evidence. In other words, we believe there is a strong case for using a margin at the high end of this range.

There are a number of issues relevant to interpreting the evidence. These fall into three groups:

• qualifications (on what the benchmarks cover);

• potential issues with the NSW market; and

• other issues in setting appropriate regulated retail tariffs.

5.1 Qualifications The benchmark entity a “mass market new entrant” is in some ways a hypothetical entity, involving a number of simplifying assumptions, which are potentially significant.

The scale of the business

We have assumed that a MMNE would need about 250,000 customers to achieve economies of scale. This would justify the investment in information systems and processes that would enable ongoing growth (eg. to 1 million customers and beyond).

What this means, however, is that in the first instance at least (see below), the relatively high fixed costs associated with operating a mass market retail business will be spread across relatively few customers (meaning higher average costs per customer). It may, however, have somewhat lower costs by virtue of having a more attractive customer base.

Time to reach the required scale

We have also assumed that a MMNE attains this scale immediately. This is, by definition, a simplistic assumption. In the event that a MMNE entered the market with the intention of competing broadly across it, then its average costs per customer would be significantly higher in the shorter term than those indicated in this report.





It is more likely, however, that any new entrant would commence operations:

• in a particular niche of the market targeting higher margin customers (and have systems to match);

• with the benefit of already competing in gas (and thus focus on dual fuel customers); and/or

• with the benefit of some generation capacity.

After securing a position, it may then attempt to grow into a truly mass market player.

It is also worth noting however that we are unaware of any instances where a stand alone retail business has entered the mass market and gained a substantial market share (at least compared to the incumbents). Making an assessment in regard to how long this process would take (even if it did happen) is therefore extremely difficult. Victoria Electricity retail business has recently grown to about 115,000 customers since 2002.40

Our assumptions, while somewhat simplistic, are nevertheless consistent with the underlying assumption regarding a mass market entrant that is able to exploit economies of scale.

5.2 Potential NSW market issues NSW is clearly the largest retail energy market in Australia and it is likely that a party with aspirations in the broader Australian mass market would enter the NSW market. That said, there are good reasons to assume that mass market new entry is likely to entail considerable risk. The key reasons are that:

• the two most likely entrants (Origin and TRUenergy) currently have limited market presence;

• there is limited vertical integration or capacity to vertically integrate (which the evidence suggests is particularly important, as we highlight below);

• the nation’s largest retailer (AGL) already has a dominant position in the retail gas market and is therefore in a strong position to offer dual fuel to the incumbent’s electricity customers. Origin and TRUenergy are less well placed;

• the remaining players in the retail electricity market are government owned, which may create a perception amongst privately owned competitors that there is a risk that the incumbents may make less commercial decisions. Some stakeholders have made this point in the context of the Energy Reform Implementation Group’s consultation process;41 and

40 This includes growing from 28,000 customers since March 2005. It has also recently made its first positive contribution to the company’s first quarter result. See http://www.infratil.com/1/6222.htm. Its owner already has a significant retail business in New Zealand. 41 Power Industry News, Edition 506, 28 August 2006, page 7.





• the market is likely to perceive a risk of the Government changing its position (eg. perhaps in response to other issues, such as petrol prices).

Vertical integration

There is a considerable body of evidence to suggest that vertical integration provides value to electricity retailers (ie. there has been considerable vertical integration in competitive retail electricity markets).

It is possible to draw inferences about the value of vertical integration by examining the differences between:

• merchant generators (or Independent Power Producers – IPPs - who typically sell their output on short term contracts – 3 years – or via spot markets); and

• contracted generators (i.e. those with long term contracts often called Power Purchase Agreements – PPAs).

The comparison between the two types of generators is instructive because the long term contracts held by the contracted generator can effectively be regarded as a proxy for the natural hedge obtained through vertical integration. By contrast, a merchant generator bears wholesale market price risk, except over the short term. These two types of generators therefore provide a useful proxy for the type of valuation differences the market attaches to the management of this risk.42

The UK provides the best available market evidence.

• In its valuation of International Power’s business, Morgan Stanley disaggregates the earnings from these two types of assets. It states:

We view the contracted earnings stream as stable and predictable… In our opinion this is worth a multiple in line with stable, regulated utilities. We apply a 15X PE multiple to gain a fair value for the contracted assets. Merchant assets should be valued differently. We believe the best way is to consider peak merchant earnings and apply a very different, much lower multiple to these. We apply a multiple of 8X to derive a fair value for the merchant assets.43

In effect, Morgan Stanley values the earnings from a merchant generator at almost half the value of those from a contracted generator. Consistent with the above, Morgan Stanley took similar issues into account in August 2004 when International Power undertook the business transforming acquisition of the non-US assets of Edison Mission Energy (which included Loy Yang B). The assets acquired where almost entirely contracted. As a result, Morgan

42 Observed differences in the valuations of these two types of generators would, however, probably overstate the benefits of achieving vertical integration in a competitive market context. This is because the vertically integrated retailer does not have the same degree of control over its market as the contracted generator, but it has greater control than a merchant generator. 43 Morgan Stanley Equity Research, ‘International Power: Further to go’, 7 March 2006.





Stanley altered its benchmark assumptions in respect of International Power’s cost of capital. For example, it altered its benchmark gearing ratio from 30% to 60%.44

Morgan Stanley also applies different discount rates when valuing pure merchant generators. For example, in respect of International Power, Morgan Stanley applies a “group” weighted average cost of capital of 7.9%. By contrast, in the cases of Drax and British Energy it uses discount rates of 8.75% and 10% respectively.45 It notes that Drax is a “higher risk investment than a typical utility because it is not vertically integrated.”46 It derives its higher cost of capital than for International Power partly by using a gearing ratio of approximately 20%.

• ABN Amro adopts a similar approach in valuing International Power. It states that:

IPR’s risk profile has changed significantly and it is not as exposed to merchant activities as it once was. Stable PPA assets now comprise 52% of IPR’s EV vs 37% in January 2004. As a result, we view IPR’s cash flows as more stable, warranting a lower WACC of 7% vs market consensus of 8%.47

It assumes IPR’s benchmark gearing ratio is 50%, based an assumed gearing of 70% for contracted assets and 30% for merchant assets. These assumptions imply a cost of capital range of about 8% to 6.2% for merchant and contracted generation assets respectively.

A similar trend to vertical integration is evident in Australia. Credit Rating agency, Fitch, recently stated that:

Integrated utilities want to strengthen the quality of their internal hedges by combining generation and retail operations through further acquisitions. The resulting combination of size, synergies and the increased degree of vertical integration is placing pressure on disaggregated entities such as state government-owned generators and retailers to consider structural ownership changes to better manage industry risks.48

A MMNE is likely to require a higher margin if it is not vertically integrated.

5.3 Other issues in setting appropriate regulated retail tariffs There are a couple of other points worth making in relation to what the benchmarks cover. These include:

• the costs to incumbents; and

• benchmarking to set cost reflective regulated retail electricity tariffs.

44 This is consistent with the actions of International Power which used 80% debt to fund the contracted assets. 45 Morgan Stanley Equity Research, ‘Drax: Confirming positive outlook’, 8 March 2006 and Morgan Stanley Equity Research, ‘British Energy – Key Takeaways from Q3 – We still prefer Drax’, 24 February 2006. Morgan Stanley uses a higher discount rate for British Energy in part because of its higher operating leverage (ie. higher fixed costs). 46 Morgan Stanley Equity Research, ‘Drax: Discounting US$43 Long-term Oil Price – May Be Conservative’, 18 January 2006, page 18. 47 ABN Amro, ‘International Power: Switching the Focus’, 13 October 2005, page 1. 48 Fitch Ratings, Australian Utilities: What’s the Mix for 2006?, 3 February 2006, page 1.





The costs of incumbents

Benchmarking the costs of a MMNE, by definition, excludes additional costs (and/or obligations) that might be borne by incumbents but not by the MMNE. There are a variety of costs that it would be reasonable to expect incumbent retailers to incur that our benchmarks do not cover. These costs are likely to include the following:

• the cost of existing legacy systems;

• the costs of a “retailer of last resort” facility (which would include additional costs for being able to quickly take over a substantial customer base and establish energy contracts to suit their demand); and

• the costs/obligations of meeting other requirement (eg. in respect of vulnerable customers).

This is unlikely to impede the ability of a new entrant to compete in the shorter term; indeed, it may help. It is, however, likely to distort the basis of competition if regulated tariffs are set too low, by focussing it more on the customers who are in a better position to avoid these costs (eg. miscellaneous charges).

Benchmarking to set regulated retail electricity tariffs

To set cost reflective regulated retail tariffs it is also important to ensure that you are valuing the benefits regulated tariffs provide compared to the competitive alternative. In other words, if regulated tariffs are to be set at competitively neutral levels, then it is crucial that the comparison is ‘like with like’. Regulators should therefore be estimating the cost of providing product features or benefits that regulated tariffs provide.

There are several areas of service differential between regulated and market offers.49 These potentially include:

• the flexible term associated with regulated tariffs;

• the reversion opportunity associated with regulated tariffs; and

• the way regulators charge for miscellaneous services.

On the first point:

• market tariffs typically involve a fixed price for a fixed term, sometimes with fees for early termination. Indeed, the fixed nature of the contract is its most fundamental attribute. Terms are typically 2-3 years and termination fees are up to $125. Alternatively, some retailers embed the risk and cost of early termination in higher unit prices. Many similar products (eg. telecoms, insurance, home loans) have similar features; and

49 We are aware that some years ago IPART concluded that there may be no service differential between competitive and standard offer contracts, but that was in the context of a review carried out under a different Terms of Reference. Moreover, further evidence has come to light since that time as indicated above.





• regulated tariffs, by contrast, typically involve a fixed price for a flexible term, with no fees for early termination. This is analogous to fixing the interest rate on your home loan while retaining the ability to switch, at no cost, to another supplier if interest rates fall.

Regulated tariffs therefore provide customers with an option to stay on fixed prices for as long as prices are set, but to move for free onto a market contract in the intervening period. A customer on a market contract does not have this luxury. Moreover, a competitive market is unlikely to provide this option, at least not for free.

The regulated offer, however, means that there is no cost to the customer of making no decision on choosing their supplier. In other words, in the event that both regulated and market offers are similar, the regulated customer retains the option and its value.

Reversion policies can compound this problem (eg. the option to leave the regulated tariff is reinforced by a further free option to return). The Minister of Energy’s Terms of Reference highlights the reversion policy and the optionality it creates.

Paul Joskow, an eminent energy economist, has noted the existence of options in the regulated retail tariffs of US Electricity Service Providers and their importance. He states that:

…allowing customers that choose to take service from an ESP to return to a regulated tariff when wholesale prices are high without being charged an appropriate price for this option, seriously undermines the development of retail competition because it effectively provides a subsidised option for retail customers who switch back and forth and a very unstable customer base for ESPs.50

Options are often quite valuable to customers, particularly when they face making decisions which involve considerable uncertainty. The retail energy purchase decision involves considerable uncertainty because the choice is quite new, the product is quite complex and customers are unlikely to devote much time to making a decision on their retailer.

To set regulated tariffs at cost reflective or competitively neutral levels, it is important that the margin incorporated into them accounts for the additional benefits they provide.

It is possible that IPART could deal with the risks and costs of service differentials in other parts of its analysis.51 However, it is not obvious that this can be done solely by reference to the evidence. For example, the evidence on which regulators rely often draws on work by other regulators, partly because the market evidence is incomplete. In addition, the market evidence tends to rely on the margins for market players who are vertically integrated in their other markets (and are better placed to manage such risks).

Despite the greater switching rates and entry into some other Australian energy markets, regulators in those jurisdictions also have yet to conclude that the markets are competitive enough to warrant deregulating, even for the majority of domestic customers in those markets.

50 P., Joskow, The Difficult Transition to Competitive Electricity Markets in the US, AEI-Brookings Joint Centre for Regulatory Studies, July 2003, page 35. 51 For example, by referring to the market evidence on what a MMNE needs to compete in other markets with similar service differentials, or by adjusting forecasting and hedging costs.





The value of the additional benefits regulated tariffs provide should be taken into account when setting a regulated retail margin, if the Minister of Energy’s primary objective is to be met.





A EnergyAustralia’s terms of reference EnergyAustralia has engaged KPMG to estimate an appropriate allowance for retail margins and operating costs to be considered by IPART when setting regulated retail tariffs.

Operating costs

We have been asked to consider the fixed and variable costs associated with (but not limited to):

• customer service;

• finance;

• information systems;

• marketing and advertising;

• customer acquisition and retention costs; and

• legal and regulatory compliance

We have been asked to consider benchmarks in other jurisdictions and internationally to the extent they may be relevant in the context of NSW regulated retail electricity tariffs.

Retail margins

We have been asked to consider an appropriate margin for a MMNE retailer, providing an appropriate return on investment having regard to the levels of:

• working capital;

• tangible and intangible assets;

• depreciation and amortisation;

• bad debts;

• competition from electricity substitutes; and

• purchase risk.

We have been asked to have regard to the effect of margin on the level of competition in a mass market.





B Operating cost benchmarking This appendix explains our approach to the development of the calculations which support our benchmark of the operating costs of a MMNE. Our:

• assumptions on the relevant business volumes are explained in section B.1.2;

• cost tree which establishes the methodology of calculating the relevant operating costs is set out in section B.2; and

• data, sources and benchmarks and explanations are set out in section B.3.

B.1 Business assumptions

B.1.1 Base year The results presented in this report are expressed in real 2005/06 Australian dollars.

Where international benchmarks are quoted, we have converted these to Australian dollars using an exchange rate relevant to the year in which the benchmark is quoted.

Where we have used benchmarks from financial years prior to 2005/06, we have inflated them to 2005/06 dollars using the weighted average CPI for 8 capital cities52 as quoted by the Australian Bureau of Statistics.

B.1.2 Volume assumptions To establish costs for a MMNE we have had to make assumptions about the size of the business that it is servicing and other issues which impact on the environment that the business operates in. Those input assumptions are set out below.

It is worth noting that some of the costs are step variable and so extrapolating them to a different sized business, or a business operating under different conditions, could be problematic.

B.1.3 Size of business We have used the following key input assumptions in our calculations.

52 We understand that this inflator is consistent with that used by IPART in regulatory determinations in NSW.





Table B - 1: Volume assumptions applied in our calculations of retail operating costs for a MMNE Input Assumption

Customer numbers 250,000

Billing intervals 245,000 customers billed quarterly

5,000 customers billed monthly

Reminder notices 30% to 40%

Call centre use Average of 2 calls per customer per year for retailer interaction

Time of use metering Installed for all of the monthly billed customers

We have chosen a size of business that warrants systems suitable to support the delivery of energy retailing to a mass market. Retailers who operate in a mass market must have systems of sufficient size to be able to handle the volumes of data and customer information in a timely manner to meet the industry code requirements and commercial customer service.

We note that there are few instances of smaller retailers serving a mass market. We do observe that:

• ACTEWAGL serves 150,000 customers, but is supported by AGL and is likely to extract some economies of scale from that affiliation;

• Aurora Energy operates in Tasmania and serves some 250,000 customers and has recently developed a custom built customer information system; and

• Victoria Electricity retail business has recently grown to about 115,000 customers since 2002.53 Its recent annual report provided no indication of any pending major IT investment.

Other energy retailing businesses are either smaller (not operating in the mass market) or operate with large customer bases of approximately 500,000 customers or more.

We have chosen a customer base of 250,000 as representative of the scale necessary to achieve the economies of scale associated with having a commercial customer information and billing system, automated business to business communication systems and energy trading systems. We base our estimate on our observations of Australian energy retailers, and in particular the size of the customer base and the systems employed.

B.1.4 Pro-forma financials Some of the benchmarks used in this report relate to data acquired from survey respondents which is classified into groups based on revenue values. Therefore, in determining the size of a MMNE, we have constructed a value for revenue and other expenses which are consistent with the volumes assumed in B.1.3 above, and average market parameters consistent with those for NSW electricity customers. Our assumptions are explained below. 53 This includes growing from 28,000 customers since March 2005. It has also recently made its first positive contribution to the company’s first quarter result. See http://www.infratil.com/1/6222.htm. Its owner already owns a significant retail business in New Zealand.





Table B - 2: Financial assumptions for a MMNE retailer Item Assumption $m

Revenue Customers 250,000

Average annual charge $3,400 for monthly billed Average annual charge $950 for quarterly billed

250.0

Energy 35% of revenue 88.0

Network charges 42% of revenue 105.0

Other costs 3% of revenue 7.5

The percentages for energy, network charges and other costs are consistent with EnergyAustralia’s experience for the year ended to June 2006, however the cost benchmarking is not sensitive to these inputs as they only serve to assist in any calculation of working capital requirements.

In the first instance, a MMNE is likely to focus on more profitable customers, which are generally larger users. However, with a customer base of 250,000 it is likely to have a broad range of customers, which will mean its customer base is close to the average customer. The incumbents are also likely to respond with competitive behaviour of their own in the most attractive customer segments.

Therefore, we consider that these outcomes listed in the assumptions above are not unreasonable for the NSW market.

B.2 Cost tree To benchmarks the costs associated with our MMNE we have created a cost tree based on the activities that we consider it will need to undertake. An overview of the cost tree is set out below.





Figure B-1: Conceptual Model and Information Requirements - Retailer Costs to Serve

2A

2B

2C

Billing and customer collection (including

CIS/ITS)$

Call centre costs$

FTE employee costs and overheads)

$

1

ADD

ADD





Figure B-2: Conceptual Model and Information Requirements - Billing and Customer Collection

3ABilling$

Data validation$

Customer transfer$

2A

ADD

ADD

Credit collection$

Bad debt expense$

CIS$

ADD

ADD

ADD

IT expense

% of revenue

Bad debt expense

% of revenue

$/receipt

# receipts

$/bill

# bills

$/transfer

% of customer transfer

# customer

Refer to section B.3.1.2









Figure B-3: Conceptual Model and Information Requirements - Billing

Figure B-4: Conceptual Model and Information Requirements - Call Centre

Reminder notices

Customer invoicing$/invoice

# invoicesRefer to section B.3.1.1

3A

Labour costs$

$/call

# calls

# FTEs

Refer to section B.3.2

2B

ADD

Overheads$

Benchmarked as 34% of total call centre costs





Figure B-5: Conceptual Model and Information Requirements - FTE Employee Costs and Overheads

Office and administration service costs

$

Energy trading$

Public relations and customer communications

$

2C

ADD

ADD

Pricing and risk management

$

Settlements$

Regulatory costs$

ADD

ADD

ADD

$/FTE

# FTEs

$/FTE

# FTEs

$/FTE

# FTEs

Advertising

$/FTE

# FTEs





Office expenses

Rental $/FTE

# total FTEs

3C





Figure B-6: Conceptual Model and Information Requirements - Regulatory Costs

Labour costs$

3C

ADD

Ombudsman scheme$

License fee$

ADD

$/FTE

# FTEs

$/customer

# customersRefer to section B.3.3.6





B.3 Data, benchmarks and calculations

B.3.1 Billing and customer collection

B.3.1.1 Billing The billing function relates to the preparation and distribution of bills to customers. The number of bills issued which in turn is a function of the number of customers and frequency of billing drives the cost of this function.

We have applied a cost of $0.73 per bill or reminder for the MMNE.54

As a further validation, the 2002 KPMG Benchmarking Study reported a tenth percentile cost of $0.36 per bill and a first quartile cost of $1.94 for the utilities and telecom industry55. The processes captured in this benchmark include the developing procedures for customer billing, maintaining customer account records, and issuing credit notes and refunds. This is a very broad range and reflects the scale of operations in the population involved in the survey. We expect that the lower end of the range refers to simple single tariff billing statements whilst the upper end is likely to include detailed billing statements and variable tariffs for telecommunications. The detailed results of the survey respondents are not known, however, the range validates the Australia Post survey benchmarks applied.

We have arrived at our cost based on the assumption that the MMNE will be despatching all customer bills by mail. We acknowledge that electronic billing will help reduce this cost. However it is in our opinion that it is rather early at this stage for this to be significant.

We have made an allowance for the issue of reminder notices to customers which will add to billing costs. The Office of Regulator General, Victoria estimated that approximately 40% of customers need reminder notices.

We acknowledge that the number of reminder notices will very much depend on the ability of a MMNE’s customers to pay their bills promptly. We expect a MMNE to be rigorous in managing its payment and credit terms, and it may therefore be able to reduce the need for the sending out of reminder notices to its customers. The uptake of direct debit as a payment method is also becoming increasingly popular, further contributing to the reduction of reminder notices. On this basis, we have suggested a lower bound benchmark of 30%, consistent with the experience of some electricity retailers in NSW.56

54 Includes recent industry estimates based on work undertaken for Australia Post for mass market issuing of paper bills to customers. 55 Industry-specific comparisons for Customer Invoicing, 2002 KPMG Benchmarking Study 56 Experience as advised by EnergyAustralia as indicative of the NSW market.





Table B - 3: Total customers and bills

Number of bills for the MMNE 2005/06 Benchmark

Customers Bills

Customer billed quarterly (#) 245,000 980,000 Customers billed monthly 5,000 60,000

Total bills 1,040,000

Table B - 4: Billing costs

Billing costs 2005/06 Benchmark

Low High

Bills issued (#) 1,040,000 1,040,000 Percentage of reminder notices 30% 40% Reminder notices – quarterly 294,000 392,000 Reminder notices - monthly 18,000 24,000

Total bills and reminders (#) 1,352,000 1,456,000 Cost per bill ($/bill) 0.73 0.73

Benchmark cost ($m) 0.99 1.06

B.3.1.2 Data validation Data validation refers to the process of checking the accuracy, validity and completeness of data used in the billing process. It will involve the investigation and correction of errors in records.

In a 2001 gross margin review for the Queensland Treasury57, we calculated that data validation costs to be in the order of $0.81 per bill. It was the Office of the Regulator General’s view that one third of the data validation costs should be allocated to the retail business where the distributor was responsible for meter reading.58

We have inflated this number to $0.93 per bill to represent 2005/06 dollars, and applied one third of this cost at $0.31 per bill as the benchmark for data validation cost for the MMNE.

57 A commercial gross margin for Queensland Retailers, June 2001, KPMG Consulting. 58 2001 Price Review, Cost allocation, September 2001, Office of the Regulatory General, page 3.





Table B - 5: Data validation costs

Data validation costs 2005/06 Benchmark

Low High

Bills issued (#) 1,040,000 1,040,000 Data validation cost per bill ($/bill) 0.31 0.31


B.3.1.3 Customer transfer Customer transfer costs are those incurred due to customers moving premises and switching to other retailers, or to a market contract with the first-tier retailer.

In a 2001 KPMG Consulting report to the ORG Victoria59, the customer transfer cost was set at $12 per transfer. We understand that the ORG accepted this cost. We have adjusted this number for inflation to arrive at $14.32 per transfer for 2005/06. Rather than applying the CPI to our original cost, we have proceeded with escalating this figure by the Labour Price Index for Sydney. The basis for doing this is the cost reflects a labour-intensive activity and inflating it using the CPI might underestimate the results as labour rates in Sydney have a tendency to increase at a relatively higher rate.

We have assumed that householders move premises once in every five years, resulting in a customer transfer rate of 20% per annum. The ORG report in 2001 estimated householders move once in every seven years, but we have acknowledged that householders are now more mobile and relocate more frequently as evidenced in a recent social study60. This is also compounded by demographic trends such as:

• the ‘sea change’ which sees the major shift of population to the coastal areas especially over the next decade as the baby boomers retire;

• people in their twenties who have been priced out of the housing market due to the property boom and therefore rent, further increasing their mobility.

In relation to business customers, we have estimated that businesses change premises (or ownership) once in every five years on average.61

We expect that customer contestability will increase in the future and customers will have greater flexibility in determining their choice of retailers and thereby increasing future transfer rates. We have allocated a further 3 and 5 percentage points to account for this for the lower and upper bound benchmark respectively, bringing the final transfer rate to 23% and 25%.

59 Ibid, page 43 60 “Changing homes is a moving thing” by Hugh Mackay, Sydney Morning Herald, January 11, 2003 61 Based on our previous work in energy retailing in another Australian jurisdiction, 2006.





Table B - 6: Customer transfer costs

Customer transfer costs 2005/06 Benchmark

Low High

Customers (#) 250,000 250,000

Transfer rate (%) 23% 25%

Customer transfers per annum (#) 57,500 62,500

Cost per transfer ($/transfer) $14.32 $14.32


B.3.1.4 Credit collection This function relates to all activities related to the collection and receipting of payments made on the bills issued, and is therefore largely driven by the number of bills raised.

The 2002 KPMG Benchmarking Study reported a first and third quartile cost of USD 0.29 and USD 5.63 respectively for the utilities and telecom industry.62 These figures were converted to Australian Dollars and then escalated by CPI, resulting in a credit collection cost per customer of $0.57 and $11.13.

The large range reflects a combination of highly efficient and less efficient companies in the coordination of the accounts receivable system. The higher end represents study participants utilising the physical settlement of bills. The processing of cheques sent by mail or cash paid over-the-counter undoubtedly will require more time and cost than when performed electronically through EFTPOS or internet banking.

We note that the NSW market is supplied a variety of payment options including higher cost and low cost transaction options such as cheques by mail, payment through Australia Post, credit card, BPay, POSTBillPay direct debit and regular payment option plans. We would expect a MMNE to offer payment systems which are efficient to the MMNE, but still comply with the industry standards required of a licensed retailer. We would not expect a MMNE to offer expensive payment options without passing the cost of that option onto the customer.

We have applied the World Class finance benchmark median cost of $2.73 per receipt to our MMNE on the assumption that there would be a mixture in the methods of payment made by its customers, and that some but perhaps not all of the less attractive options would be offered to attract the mass market.

62 Industry-specific comparisons for Accounts Receivable, 2002 World Class Finance Benchmark Results.





Table B - 7: Total receipts

Number of receipts for the MMNE 2005/06 Benchmark

Customers Bills

Customer billed quarterly (#) 245,000 980,000 Customers billed monthly 5,000 60,000

Total receipts 1,040,000

Table B - 8: Credit collection costs

Credit collection costs 2005/06 Benchmark

Low High

Total receipts (#) 1,040,000 1,040,000 Cost per receipt ($/receipt) $2.73 $2.73


B.3.1.5 Bad debt expense We have defined bad debt expense as the net write-off of accounts after recoveries. It does not include costs associated with reminder and disconnection notices. Reminder notices are covered in the billing process, and disconnection notices and disconnection actions are generally charged to and recovered directly from the relevant customers. It is also worth noting that a reminder notice does not necessarily suggest that the customer is a credit risk. Reminder notices are system produced and may occur at the same time or shortly before the receipt is processed.

The 2002 KPMG Benchmarking Study provided a benchmark for the bad debts incurred by companies in a variety of different industries.63 At the first quartile, the percentage of bad debt write-offs to sales is 0.060%. This is relatively low and reflects industries other than electricity retailing which is more highly regulated with regard to the provision of an essential service. The 2002 KPMG Benchmarking Study quotes a median for bad debt expense as 0.327%.

Our analysis of electricity (or energy) retailers as outlined in the following table suggests that the bad debt expense rates experienced in Australia is coincidentally in the order of 0.327% for the 2004/05 year. Based on this, we estimate a bad debt expense of 0.327% in line with the industry standards for mass market retailers.

63 Comparisons for Percentage of Bad Debt Write-off to Sales, 2002 World Class Finance Benchmarks.





Table B - 9: Comparison of bad debt for electricity retailers and distributors

Retailer/Distributor 2004/05 Data

Revenue ($m) Bad debt ($’000) Percentage Citipower/Powercor 941 1,212 0.13 Origin Energy 4,931 16,107 0.33 AGL 4,915 20,400 0.42 Integral Energy 1,369 6,353 0.46 Energy Australia 2,836 9,100 0.32 Average of above data 0.327%

The above data was extracted from Annual Financial Statements for the 2004/05 financial year.

Table B - 10: Bad debt expense

Bad debt expense 2005/06 Benchmark

Low High

Total sales or revenue ($m) 250 250 Percentage of bad debt 0.33% 0.33%


B.3.1.6 Customer information system costs We have defined CIS as the systems necessary for an MMNE to undertake its retailing business activities which includes:

• trading and settlements;

• customer relationship management systems (including call centre requirements);

• data warehousing;

• forecasting and marketing software;

• maintenance of hardware;

• IT personnel costs and overheads; and

• IT training and license fees.

Further, CIS and ITS costs also include annual personal computer costs for the corporate and administration functions, trading, settlements, regulatory, marketing or pricing, and the CIS requirements of the call centre (as described above).





A 2004/05 KPMG Utilities IT benchmarking survey64 showed that the average adjusted IT operating expenditure as a percentage of corporate revenue was 1.7%. The term ‘adjusted’ refers to the exclusion of costs which had not been taken at a corporate level. Examples of these are the depreciation of IT assets and occupation overheads.

Table B - 11: Customer information system cost

CIS cost 2005/06 Benchmark

Low High

Total sales or revenue ($m) 250 250 Percentage of IT operating expenditure 1.7% 1.7%


We note that in NSW and Victoria in particular, there is an increase in the roll out of time of use metering amongst smaller consumers. This has the potential to impact on costs in the future due to the increased amount of data handled, tariff complexity and data validation costs.

NSW (and Victoria) is early in the process of increasing the penetration of time of use metering. As a result, there is a significant degree of uncertainty over what the CIS cost impacts might be and how they might impact on the non-incumbent retailers. In part, this will depend on how they respond to the roll out of interval meters.

We have not therefore costed this factor into the benchmarks, but note that these costs are likely to show an upward trend over time.

B.3.1.7 Total for billing and customer collection The total cost for billing and customer collection is summarised below.

Table B - 12: Total benchmark cost for billing and customer collection

Benchmarks 2005/06 Benchmark

Low High

Billing 0.99 1.06 Data validation 0.32 0.32 Customer transfer 0.82 0.90 Credit collection 2.84 2.84 Bad debt expense 0.82 0.82 CIS 4.25 4.25

Total for Billing and customer collection (including CIS/ITS) ($m) 10.04 10.19

64 The study into IT costs of Australian and New Zealand utilities had 12 participants and covered the period of January 2005 to July 2005. The information used was for financial year ending 2004.





B.3.2 Call centre A MMNE will require a contact centre to administer incoming customer product support or information inquiries. This is a principle activity of a mass market retailer where the services are delivered through a third party supply network rather than a “shop front” interface.

We benchmark the costs of operating a call centre by examining the number of calls attended to by the call centre. The number of calls is likely to be a function of:

• certain aspects of the retailer’s communications, e.g. a clear and concise invoice design will reduce customer confusion and therefore, the number of calls made by customers; and

• industry activities or campaigns, e.g. the push by the government for more flexibility in energy choices might result in the rolling out of more energy plans, thereby increasing the number of calls asking for more information.

Publicly available information shows that in South Australia energy retailers received a total of 1.396 million calls in 2002/03 from 0.740 million customers.65 This suggests that a customer makes approximately 1.9 calls per annum. In Victoria we found find that a total of 2.278 million customers made 4.715 million calls to customer centres in the same year, or an average of 2.1 calls per customer.66

We have chosen to use the figures for 2002/03 because we consider that the data collected during this time is more accurate than the available data for subsequent years. As contestability in the electricity market expands, it has become increasingly difficult for call centres to disclose data on the origin of callers, especially so when the electricity retailer operates a national call centre servicing retailing operations in multiple states.

We have estimated that a MMNE’s call centre would receive an average of 2 calls per customer in a year. Further, we have assumed that call centre agents can respond to 80 calls per day (12 calls per hour for 6.5 hours). The number of full-time equivalent (FTE) agents required to handle this volume determines the labour cost for the call centre.

We have applied the following ratios of contact centre staff as recommended in an ACA research67 paper. This suggests that the call centre would need approximately 28 FTEs68 for its operations, resulting in a call centre labour cost of $1.35m.

The ACA research benchmark also suggests that labour costs comprise 66% of total call centre costs (including the CIS). Hence, the total call centre cost for the MMNE is $2.05m.

65 http://www.escosa.sa.gov.au/webdata/resources/files/050401-R-SCONRRRetailTemp03_04.pdf 66 http://www.esc.vic.gov.au/NR/rdonlyres/D8B79DE7-D3E9-4B4B-AE86-77F6F6FC70CB/0/NationalTemplate2004_05.pdf 67 ACA research, 2002/03 and 2003/04 National regulatory requirements – Comparative summary for retailers to small electricity customers in Queensland, Office of Energy 68 A total of 1 managers ($150,000), 3 team leaders ($65,000), 22 agents ($42,000) and 2 support staff ($42,000). All salaries for contact centres as benchmarked by Hays Salary Survey 2006. These salaries are at the higher end of the range consistent with the assumption that the MMNE will have a large contact centre to serve a mass market.





Figure B-7: Call centre staff ratio

We are aware that certain technologies applied by call centres could affect the call centre costs to callers. For example, interactive voice response (IVR) allows the caller to select an option from a voice menu to connect customers to their desired service promptly. It is not unreasonable to assume that automated features have increasingly been used at contact centres these days. Our benchmarks reflect the current state of automation69 but note that this may change somewhat over time with improved technology, and with changes in the level of customer acceptance.

Table B - 13: Call centre costs

Call centre costs 2005/06 Benchmark

Low High

Labour costs ($m) 1.35 1.35 Percentage of labour cost to total call centre cost (%) 0.66 0.66

Total for call centre cost ($m) 2.05 2.05

B.3.3 Full time equivalent employee costs and overheads Corporate and administrative overheads are:

• salary and wages costs associated with the management and administration of the business entity; and

• leasing and maintenance costs related to the provision of setting in which the retailer trades.

We have grouped the various staff into several key groups, namely:

69 Our benchmarks were collected in 2003, and reflect the current levels of automation employed in call centres. Costs may change as more interactive systems are developed and as technology improves in communication management however the impact is not likely to be significant in the immediate future.

Agents

Team leader

Call centremanager

1

2.5

22.5

Support staff 1.25





• corporate and administrative;

• energy trading;

• public relations and customer communication;

• pricing and risk management;

• settlements; and

• regulation and compliance.

We have identified the number of FTE employees required for each corporate and administrative role and their respective average salaries as suggested by the 2006 Hays Salary Survey. All quoted numbers are for Sydney, NSW. They are representative of the total package value including superannuation.70

In our suggested staff structure, we have not made an allowance for IT or call centre staff for data validation and customer transfer activities. These are accounted separately in sections B.3.1.6 and B.3.2. The last column on the right of the table provides a reference to the specific section of this report for a more detailed discussion.

Table B - 14: Staff breakdown

Position Lower bound FTE

Upper bound FTE

Salary per FTE ($m) Reference

Corporate and administrative roles CEO/Managing Director 1 1 0.45071 Personal assistant 1 1 0.062 Company secretary 1 1 0.057 Board of director (number engaged in corporate governance, not FTE positions) 4 5 0.06872

Legal counsel 1 2 0.100 Corporate and strategic planning 1 2 0.100 Human resources manager 1 1 0.110 Human resources officers 1 1 0.065 Payroll 1 1 0.058 Chief financial officer 1 1 0.250 Corporate risk management 1 1 0.140 Finance manager 1 1 0.160 Finance, admin. and business services officers 2 2 0.080

70 Except for personal assistant and secretary where salaries were representative of the cash component only, but lifted by 15% to reflect superannuation and other costs. The package also includes bonuses and other benefits. 71 We have used the remuneration package for Envestra Limited as a guide and scale according to the size of the MMNE. Source: Envestra Limited’s annual report 2005 We have then adjusted the rate to reflect a Sydney package as compared to an Adelaide package with reference to the differential for CFO as quoted by Hays Salary Survey.





Position Lower bound FTE

Upper bound FTE

Salary per FTE ($m) Reference

Tax manager 1 1 0.140 Treasury officers 1 1 0.086 Administrative staff and reception 2 2 0.044

Subtotal for corporate and administrative roles 21 24 Section B.3.3.1

Energy trading Manager, trading 1 1 Senior traders 1 2 Energy traders 0 1 Market analysts 1 1

0.14072

Subtotal for energy trading 3 5 Section B.3.3.2

Public relations and customer communication Marketing professionals 1 1 Public relations manager 1 1

0.096

Customer relations officers 2 2

Subtotal for public relations and customer communications 4 4 Section B.3.3.3

Pricing and risk management Pricing manager 1 1 Data analysts (pricing) 2 2 Risk manager 1 1 Risk analysts 1 1

0.120

Subtotal for pricing and risk management 5 5 Section B.3.3.4

Settlements Settlements manager 1 1 Settlements analysts 2 2

0.100

Subtotal for settlements 3 3 Section B.3.3.5

Regulation and compliance Regulatory manager 1 1 Compliance manager 1 1 Regulatory analysts 1 1 Compliance analysts 1 1

0.135

Subtotal for regulation and compliance 4 4 Section B.3.3.6

Total staff and total incremental staff 40 45

B.3.3.1 Office and administration service costs The extent of investment needed for office infrastructure is a function of the number of people in the workforce.

72 Based on KPMG’s previous work for an electricity distributor and retailer, 2006.





The breakdown in the above table suggests that the MMNE would have 40 to 45 FTEs, including a board of directors. We have based this structure on our knowledge of the corporate and administrative function of other retail businesses.

The annual cost of $22,283 per FTE is derived from:

• office leasing costs of $17,910 per annum. We estimated that an FTE would require approximately 30 square meters of working space, and multiplied this by the average Sydney CBD corporate office rates for May 2006 of $597 per square meter73; and

• a suggested loading factor of 30 percent for the expenses incurred for utilities, maintenance and taxes.

We have estimated a sum of $0.565m to $0.60074 for the following additional office expenses:

• postage and freight;

• printing and stationery;

• telephone and facsimile;

• travel;

• consultancy expenditures; and

• sponsorship and corporate entertainment costs.

Table B - 15: Corporate and administrative staff

Corporate and administrative personnel 2005/06 Benchmark

Low High

FTE (#) 21 24

Total ($m) 2.19 2.46

This table represents the corporate and admin staff identified in table B-14 above. The labour and on-costs associated with other staff involved in billing, energy trading etc. are covered in the respective area.

73 Office rents and occupancy costs, CB Richard Ellis Global Market Rents, May 2006 74 Based on previous work for an Australian distributor and retailer, 2006.





Table B - 16: Office and administration service costs

Office and administration service costs 2005/06 Benchmark

Low High

FTE (from table B14 above) (#) 40 45 Rental and utilities costs per person ($m) 0.022 0.022

Subtotal – rent and utilities ($m) 0.88 1.00 Office expenses, travel, consultants, postage and freight ($m) 0.57 0.60

Total rent, utilities and office expenditure ($m) 1.45 1.60 Add Salaries and wages for Corporate admin from Table B-15 above ($m) 2.19 2.46

Total ($m) 3.64 4.06

B.3.3.2 Energy trading The energy trading function of a MMNE is responsible for managing the risks of its exposures in the physical and financial markets.

We base our estimates upon similar engagements undertaken in the past.75 We consider that an energy trading room for a MMNE would have the staff shown in Table B-17 below.

We estimated that a small energy trading function would require 2 to 3 FTEs with a cost of $0.420 million per annum. This cost could increase to $0.700 million for a larger trading room with a 24-hour operation and require 5 to 7 staff.

The MMNE would likely need a trading function of significant size to cope with its customer base and to implement appropriate strategies, particularly with increased trading risk following greater contestability and the removal of ETEF.

Table B - 17: Energy trading costs

Energy trading costs 2005/06 Benchmark

Low High

FTE (#) 3 5

Total ($m) 0.42 0.70

B.3.3.3 Public relations and customer communications Customer communication costs include those associated with informing customers of:

• available payment options;

75 Based on previous work for an Australian distributor and retailer, 2006.





• their rights under an applicable retail code or customer charter;

• the availability of franchise (price regulated) green power options;

• the connection and disconnection process applying to the franchise market; and

• the cost of ensuring that the franchise retailer’s brand is sufficiently distinguished from any related business (such as distribution), thus meeting ring fencing obligations.

As part of the review of franchise gross margin for a retail electricity business in another jurisdiction, we have estimated this cost to be in the vicinity of $2.5 to $3.0 million. It is reasonable to draw a comparison between this retailer and the MMNE as both of them have a customer base across a similar geographical region, and would use the same media to communicate with these customers. For example, we observe that another retail electricity business regularly advertises in one or two major newspapers in New South Wales as retailers appear to view this as one of the most effective way of communicating with the target market.

It is also our experience that advertising and marketing costs are not likely to vary much with volumes, especially for a large retailer approaching its customer base with a mass market advertising campaign.

Table B - 18: Advertising and marketing

Advertising and marketing 2005/06 Benchmark

Low High


Further, there are also the costs from salaries and wages attributed to this group. Table B-19 below outlines the total cost.

Table B - 19: Public relations and customer communication costs

Public relations and customer communication costs 2005/06 Benchmark

Low High FTE (#) 4 4 Salaries and wages ($m) 0.38 0.38 Advertising and Marketing ($m) 2.50 3.00

Total ($m) 2.88 3.38

B.3.3.4 Pricing and risk management The pricing and risk management team is required to mitigate the risks brought by the exposure to the volatility of prices and volumes. Specifically its roles are to:

• develop franchise tariffs consistent with the imposed regulatory constraints;





• identify risks to the business generated by different tariff categories;

• analyse market data to advise management on emerging issues and business impacts; and

• assess the impact of customer switching on appropriate tariff structures for residual franchise electricity customers.

We suggest a team as outlined below.

Table B - 20: Pricing and risk management staff

Pricing and risk management

FTE (#) 5 Total ($m) 0.6

B.3.3.5 Settlements The suggested number of FTEs and labour costs associated with the settlements function is summarised in Table B-21 below. This function entails energy trading, settlements, reconciliations, verification and contracting.

Table B - 21: Settlements staff

Settlements

FTE (#) 3

Total ($m) 0.3

B.3.3.6 Regulatory costs Regulatory costs are those incurred in complying with regulatory requirements. We have included in the regulatory costs of a MMNE the following:

costs of retail license;

• costs of participation in the energy ombudsman scheme; and

• costs of regulatory and compliance personnel.

License fee

The retail license fee is estimated to be $0.0523m per annum based on IPART published data76 .

76 Based on annual electricity license fees for 2003/04 as published by IPART, and escalated to 2005/06 dollars. License fee comprises of a fixed component of $10,000 plus $4,000 for every 1 percent of market share (in 2003/04 dollars).





Participation in the energy ombudsman scheme

The NSW electricity industry ombudsman is the Energy and Water Ombudsman of NSW (EWON). EWON deals with disputes and complaints arising under any other standard form customer contract.

In developing a per customer benchmark ombudsman cost incurred by the MMNE, we have examined similar rates from the Energy and Water Ombudsman Scheme of Victoria (EWOV) and the Electricity Industry Ombudsman of South Australia (EIOSA). Table B-22 shows our calculations.

Table B - 22: Comparison between costs of ombudsman schemes

Scheme Financial year Operating costs Customers covered

Number of customers Cost per customer

EWOV 2005 4.77 $0.80

2004 4.46

Electricity, gas and

water

6.0m77 $0.74

EIOSA 2005 0.85 $0.77 2004 0.77

Electricity and gas

1.1m78 $0.70

EWON 2005 3.80 $0.55

2004 3.58

Electricity, gas and

water

6.90m79 $0.52

The table shows that per unit ombudsman cost range from $0.52 to $0.80. It is worth noting that we have assumed that a household or business which is simultaneously an electricity, gas and water customer is considered three customers consistent with the opportunity for the customer to contact the ombudsman.

Based on the above, we have estimated the ombudsman scheme costs at $0.52 per customer per annum representing the NSW experience, which is at the lower end of the range of the three states examined. We expect that the MMNE will have new systems and attempt to secure slightly better than the average customer, which should result in slightly lower than average interaction with an ombudsman. We have therefore have chosen the lower “2004” NSW benchmark in our calculations, however the difference is not particularly material to the final figures. This results in a benchmark scheme cost of $0.13m for 2005/06.

It is likely a MMNE would also have customer relations personnel who are responsible for dispute resolution internally before forwarding the necessary cases to EWON. We have accounted for this additional FTE cost as part of public relations and communications.

77 Total of 2.3m electricity customers, 2.2m water customers (assuming 0.1m of electricity customers not connected to mains water supply) and 1.5m gas customers. Source: ESC – “Final report to Minister – Special investigation: Review of effectiveness of retail competition and consumer safety in gas and electricity’, 2004. Electricity and gas customer numbers for Victoria were derived from rates of consumer switching. 78 Total of 0.75m electricity customers and 0.36m gas customers. Source: KPMG industry knowledge. 79 NSW and ACT have a total of 4.262m electricity and gas customers. 6.905m customers for NSW is derived by subtracting 4.262m by 0.257m customers in ACT, and added to an estimated 2.9m water customers in NSW. We have estimated that there are 2.9m water customers. Source: ESAA Report 2005 and ActewAGL Annual Report 2005.





Regulatory and compliance personnel

The final component of regulatory costs is the wages and salaries of regulatory and compliance personnel dealing with regulatory reporting and licence compliance issues, regulatory submissions, and industry monitoring. Table B-23 shows this cost.

Table B - 23: Regulatory and compliance staff

Regulatory and compliance personnel

FTE (#) 4

Total ($m) $0.54

Total regulatory cost

The total regulatory cost for the MMNE is the sum of the three components discussed previously. It is estimated that this figure will be approximately $0.75m.

Table B - 24: Regulatory costs

Regulatory costs 2005/06 Benchmark

Low High

Total customers (#) 250,000 250,000 Ombudsman cost per customer ($) 0.52 0.52

Subtotal ($m) 0.13 0.13 License costs 0.05 0.05 Regulatory and compliance personnel 0.54 0.54

Total ($m) 0.72 0.72

B.3.3.7 Total for FTE employee costs and overheads The total cost relating to employee and overheads are detailed below.

Table B - 25: Total benchmark cost for employee costs and overheads

Area 2005/06 Benchmark

Low High

Office and administration service costs 3.64 4.06 Energy trading costs 0.42 0.70 Customer communications 2.88 3.38 Pricing and risk management 0.60 0.60 Settlements 0.30 0.30 Regulatory costs 0.72 0.72

Total ($m) 8.57 9.77





B.4 Summary The retail operating cost for a MMNE consists of all the cost-to-serve components discussed in this appendix and summarised in Table B-26 below. We have suggested a lower and upper range for the retail operating cost of a MMNE. These benchmarks were derived by identifying the key activities required of an efficient MMNE operating in New South Wales. We have where possible used the most recent cost benchmarks and escalated accordingly.

It is important to note that implicit in our cost estimates are some assumptions that we have made relating to the operations of a MMNE, as set out in the body of this report. If different assumptions are applied then it should be appreciated that different results may be produced.

Table B - 26: Summary of retail operating cost

2005/06 Benchmark

Section

Lower bound benchmark

($m)

Upper bound benchmark

($m) %

Billing and customer collection (including CIS/ITS) B.3.1 10.04 10.19 46-50 Call centre costs B.3.2 2.05 2.05 9-10 Full time equivalent employee costs and overheads B.3.3 8.57 9.77 41-44

Total 20.70 22.00 100%

The total retail operating cost for the MMNE ranges from $20.7m to $22.0m, resulting in a per customer cost of $82.65 to $88.01 for 2005/06.

B.5 Comparative analysis We have summarised the results of recent regulatory decisions on the establishment of retail operating costs in Table B-27 below. Appendix C provides further details on these decisions.

Table B - 27: Summary of retail operating cost

Jurisdictional Benchmarking Energy market

Year Operating costs per customer

Customer base ‘000

SA Electricity 2005 84 700 NSW Electricity 2004 70 Various Victoria Electricity 2003 90 Various SA Electricity 2003 82 700 Tasmania Electricity 2003 77 250 ACT Electricity 2003 85 150 NSW Electricity 2002 45 – 75 Various SA Electricity 2002 80 700 Victoria Electricity 2001 50 – 80 Various NSW Electricity 2000 40 – 60 Various





The table needs to be read in the context of the introduction of retail contestability. The lower retail operating costs per customer in the earlier years do not reflect the costs of systems required to manage energy customers in a contestable market. Contestability increases complexity of data management and information transfer between retailer and distributor, whilst maintaining customer privacy.

The number of customers in the energy market evaluated also influences the cost per customer. The range of costs in NSW and Victoria recognise the value of fixed costs and their application against a range of business sizes operating in those markets.

Considering the above, and examining the operating costs allowed for each customer we observe that regulators tend towards a range of $80 to $90 for energy retailers with less than 500,000 customers.

We note in our assumptions that a MMNE retailer will need to be of sufficient size to justify an investment in systems with capacity to enter the mass market, and have assumed a customer base of 250,000 to represent this size.

The Independent Competition and Regulatory Commission (“ICRC”) in the ACT states that the retailer operating costs sufficient to cover the costs of a business the size of ACTEWAGL are in the order of $85 per customer in 2003.80 This would be over $90 per customer in 2005/06 dollars. The ICRC notes that this is at the upper end of the range due to the diseconomies of scale as compared to Victorian and NSW or SA retailers with typically more that 500-600,000 customers.81 Therefore, we would expect that a MMNE retailer would experience operating costs at the higher end of the range due to the size of its customer base.

80 ICRC, Investigation into Retail Prices for Non-contestable Electricity Customers in the ACT, May 2003, page 22. 81 We observe that a recent market presentation by AGL on the AGL/Alinta merger suggests AGL is currently experiencing operating costs of $91 per customer, but is expecting to reduce this significantly (perhaps to $68 per customer) with the amalgamation of systems and an operating cost efficiency project to commence in the near future. AGL, A new AGL scheme booklet release (revised), 29 August 2006, page 26.



53

C Recent regulated electricity retail tariff decisions The table below describes recent net and gross margins (where applicable) and retail operating cost decisions by regulators in Australian and overseas jurisdictions for electricity retailers.

Table C - 1: Recent regulatory precedent for franchise electricity retailers- retail operating costs, net margin and RGM

Jurisdiction, Regulator and Retailer(s)

Operating cost $ per

customer

Net margin

% of turnover

Gross margin

% of turnover

Date Source Notes

South Australia (ESCOSA)

AGLRE $8482 10%83 2005 ESCOSA

The Essential Services Commission of South Australian (“ESCOSA”) decided upon a retail operating cost of $84 in December 2004 dollars, increasing by CPI + 2% thereafter. In addition, a margin of 10% on wholesale and retail operating costs, rather than a net or gross margin. The 10% retail margin applied by ESCOSA is the return on the combined wholesale electricity and retail operating costs of AGL SA, rather than a percentage return on sales. ESCOSA noted that this margin equated to around 5% of total costs given that network costs (on which no return was granted), comprise around 50% of retailer costs.

NSW (IPART) EnergyAustralia, Country

Energy, Integral Energy,

Australian Inland Energy

$7084 2%85 2004 IPART

IPART adopted a different approach than in previous years by setting target delivered tariffs for customers consuming less than 160MWh per annum, rather than benchmarking retailer margins. IPART introduced the concept of ‘N’ and ‘R’ components of delivered energy costs (network and retail costs). The decision referred to here affected the ‘R’ component only. The retail operating cost was considered largely fixed.

Victoria (ESC/Department of Infrastructure) Origin, AGL, TXU

$9086 5% 2003 Charles

River Associates

CRA suggested that $90 per customer was consistent with retailers’ views on the retail operating costs. There is no evidence that DOI accepted this view in the CPI-X tariff path adopted for 2004-07, and this retail operating cost figure was not granted by the ESC.

82 ESCOSA, ‘Final report – Inquiry into electricity retail price path’, 2005, page 53. 83 Ibid., page 57. 84 IPART, ‘NSW electricity regulated retail tariffs 2004/05 to 2006/07- Final report and determination’, 2004, page 10. http://www.ipart.nsw.gov.au/files/Det04-1.pdf 85 Ibid., page 9. 86 Charles River Associates, ‘Electricity and gas standing offers and deemed contracts (2004-07)’, 2003, page 25.



54



customer

Net margin

% of turnover

Gross margin

% of turnover

Date Source Notes

South Australia (ESCOSA)

AGLRE $8287 5%88 2003 ESCOSA

Tasmania (OTTER) Aurora Energy

$76.6789 3%90 2003 OTTER Retail operating cost is expressed in May 2003 dollars.

ACT (ICRC) ActewAGL

$8591 5% 2003 ICRC The ICRC noted that ActewAGL could not take advantage of scale economies that might be available to other interstate retailers.

NSW (IPART) EnergyAustralia, Integral

Energy, North Power, Advance Energy, Great

Southern Energy, Australian Inland Energy

$45 - $7592

1.5% - 2.5%93 2002 IPART

South Australia (SAIIR) AGLRE

$8094 2002 SAIIR/ESCOSA

Victoria (ORG) Citipower, United Energy, Solaris, Powercor, TXU

$50 - $8095

2.5% - 5%96 2001 ESC/ORG

87 ESCOSA, ‘Electricity industry guideline no. 10 – Electricity retail price justification’, December 2003, page 10. 88 Ibid., page 10. 89 Office of the Tasmanian Energy Regulator, ‘Investigation of prices for electricity distribution services and retail tariffs’, 2003, page 155 90 Ibid., page 168. 91 Independent Competition and Regulatory Commission, Report 5 of 2003 - ‘Investigation into retail prices for non-contestable electricity customers in the ACT’, May 2003, page 21 92 IPART, ‘Mid-term review of regulated retail prices for electricity to 2004’, June 2002, page 5. 93 Ibid., page 5. 94 South Australian Independent Industry Regulator, ‘Electricity retail price justification: Final report’, September 2002, page 15 95 Office of the Regulator General, ‘Special investigation – Electricity retailer’s proposed price increases – Final report’, 2001, page 33. 96 Ibid., page 33.



55



customer

Net margin

% of turnover

Gross margin

% of turnover

Date Source Notes

NSW (IPART) EnergyAustralia, Integral

Energy, North Power, Advance Energy, Great

Southern Energy, Australian Inland Energy

$40 - $6097

1.5%- 2.5%98 2000 IPART

Tasmania (OTTER) Aurora 1.5%99 1999 The regulator made reference to the low risk facing Aurora’s retailing

activities, and assumed annual productivity gains of 5%.

97 IPART, ‘Regulated Retail Prices for Electricity to 2004’, 2000 98 Ibid. 99 OTTER, ‘Investigation into electricity pricing – final report’, 1999, page 169.

5 A P P E N D I X B – A D D E N D U M T O K P M G R E P O R T F O R E N E R G Y A U S T R A L I A O N N E T M A R G I N U S I N G C O S T B U I L D - U P A P P R O A C H

ABCD

EnergyAustralia

Addendum to Benchmarking Retail Operating Costs and

Margins A calculation of the return on assets for a mass market new

entrant retailer as defined in our report to EnergyAustralia dated

October 2006

November 2006 This report contains 18 pages

EA06-RetBchMkAddGH1117f.doc

© 2006 KPMG, an Australian partnership and a member firm of the KPMG network of independent member firms affiliated with KPMG International, a Swiss cooperative. All rights reserved. The KPMG

logo and name are trademarks of KPMG. Liability limited by a scheme approved under Professional Standards Legislation.

EA06-RetBchMkAddGH1117f.doc - 5 December 2006


Addendum to Benchmarking Retail Operating Costs and Margins

November 2006

i © 2006 KPMG, an Australian partnership and a member firm of the KPMG network of independent

member firms affiliated with KPMG International, a Swiss cooperative. All rights reserved. The KPMG logo and name are trademarks of KPMG.

Contents

1 Executive Summary 2 1.1 Background 2 1.2 Findings 2 1.3 Disclaimer 3

2 Background information 4 2.1 Base year 4 2.2 Volume assumptions 4 2.2.1 Size of business 4

3 Retail margins 5 3.1 Return on capital 5 3.2 Conclusion 7

A Estimating capital employed 8 A.1 Investment in working capital 8 A.2 NEMMCO prudential guarantee capital 10 A.3 Fixed assets 11

B Estimating the cost of capital 13 B.1 Introduction 13 B.2 WACC definition 13 B.3 Capital structure (D/V and E/V) 13 B.4 CAPM cost of equity 14 B.5 Value of imputation credits (gamma) 16 B.6 Corporate tax rate 16 B.7 Pre-tax cost of debt 16 B.8 Summary 17




November 2006

2© 2006 KPMG, an Australian partnership and a member firm of the KPMG network of independent

member firms affiliated with KPMG International, a Swiss cooperative. All rights reserved. The KPMG

1 Executive Summary

1.1 Background The Independent Pricing and Regulatory Tribunal (“IPART”) is currently engaged in a process to determine regulated electricity retail tariffs for the period 2007 to 2010. As part of this process, IPART recently published an Issues Paper which calls for submissions from interested parties on the issues involved in setting regulated tariffs.1

EnergyAustralia (“EA”) has engaged KPMG to benchmark the:

• operating costs of a Mass Market New Entrant (“MMNE”); and

• retail margins a MMNE might reasonably expect to earn.

The operating cost benchmarking and the market analysis of margins are covered in a separate report titled “Benchmarking Retail Operating Costs and Margins” and dated October 2006 (“the main report”).

This document is an addendum to that report and should be read in conjunction with it. This reports deals with the calculation of a margin based on the returns applied to the capital necessary for MMNE retailer to invest in systems, working capital and the investment in marketing to new customers.

1.2 Findings The results of our calculation of the returns necessary for a MMNE retailer with revenues of approximately $250m per annum suggest a return based on a margin of between 4.3% and 7.1% will compensate the business for the necessary investment in the capital required to operate a retail business of this size in the NSW electricity market. Critical to the margin calculation is an amount for the costs in initial marketing investments to secure 250,000 customers. Our calculations have assumed an investment of $150 per customer. An alternative assumption on the investment cost per customer will necessarily result in a different required margin.

Extrapolation

Our findings relate to the MMNE as defined in section B.1 of the main report. This represents our estimate of the size of business in a MMNE retail operation. Extrapolation of the values prepared in this report to an alternate sized business may not be appropriate.

1 IPART, Review of Regulated Retail Tariffs and Charges for Electricity 2007 to 2010, July 2006.




November 2006



1.3 Disclaimer

Inherent Limitations

This report has been prepared as outlined in section 2 of the main report. The procedures outlined in this report constitute neither an audit nor a comprehensive review of operations.

No warranty of completeness, accuracy or reliability is given in relation to the statements and representations made by, and the information and documentation provided by, EnergyAustralia consulted as part of the process.

KPMG have indicated within this report the sources of the information provided. We have not sought to independently verify those sources unless otherwise noted within the report.

In the course of our work, projections have been prepared on the basis of assumptions and methodology which have been described in our report. It is possible that some of the assumptions underlying our projections may not materialise. Nevertheless, we have applied our professional judgement in making these assumptions, such that they constitute an understandable basis for estimates and projections. Beyond this, to the extent that certain assumptions do not materialise, then it must be appreciated that our estimates and projections of achievable results will vary.

KPMG is under no obligation in any circumstance to update this report, in either oral or written form, for events occurring after the report has been issued in final form.

The findings in this report have been formed on the above basis.

Third Party Reliance

This report is solely for the purpose set out in section 1 of this report and for EnergyAustralia’s information which includes the use of this information in EnergyAustralia’s response to the IPART Issues Paper. It is not however, to be used for any other purpose or distributed to any other party without KPMG’s prior written consent.

This report has been prepared at the request of EnergyAustralia in accordance with the terms of KPMG’s engagement letter/contract dated 26 July 2006. Other than our responsibility to EnergyAustralia, neither KPMG nor any member or employee of KPMG undertakes responsibility arising in any way from reliance placed by a third party on this report. Any reliance placed is that party’s sole responsibility.




November 2006



2 Background information

2.1 Base year The results presented in this report are expressed in real 2005/06 Australian dollars.

Where international benchmarks are quoted, we have converted these to Australian dollars using an exchange rate relevant to the year in which the benchmark is quoted.

Where benchmarks have been obtained from financial years prior to 2005/06, we have converted them to Australian dollars at the prevailing rate and inflated them to 2005/06 dollars using the weighted average CPI for 8 capital cities2 as quoted by the Australian Bureau of Statistics.

2.2 Volume assumptions To establish costs for a MMNE retailer, we have had to make assumptions about the size of the business that the retailer is servicing and other issues which impact on the environment that the business operates in. Those input assumptions are set out in the body of the main report. However, in recognising the size of the business assumed for this report, it should be noted that some of the costs are step variable, and the final values for the relevant costs may not necessarily be able to be extrapolated to a different sized business, or a business operating in a different market under different conditions.

2.2.1 Size of business The input assumptions identified below have been applied to our calculations and discussed in more detail in Appendix B to the main report, and summarised below.

Table 1: Financial assumptions for a MMNE retailer

Item Assumption $m

Revenue Customers 250,000

Average annual charge $3,400 for monthly billed Average annual charge $950 for quarterly billed

250.0

Energy 35% of revenue 88.0

Network charges 42% of revenue 105.0

Other costs 3% of revenue 7.5

2 We understand that this inflator is consistent with that used by IPART in regulatory determinations in NSW.




November 2006



3 Retail margins To benchmark retail margins through the application of a cost of capital to the necessary investment incurred by a MMNE we:

• undertook a cost build-up to determine the asset base necessary for a MMNE retailer to undertake electricity retailing operations in NSW; and

• multiplied this by a weighted average cost of capital (“WACC”) to the asset base. The WACC is calculated based on various assumptions outlined in this report.

3.1 Return on capital The MMNE retailer will need to invest in working capital and fixed assets and requires a return in order to finance an operation that seeks to supply electricity customers. Much of the working capital requirements are driven by industry parameters and regulation which are outside of the control of the MMNE. We have therefore established a calculation based on our analysis of the market conditions in which we have assumed the MMNE will operate.

We establish a benchmark asset value by determining:

• a benchmark of franchise retail business net working capital;

• prudential capital required for energy purchases; and

• a benchmark for retail business fixed assets.

This benchmark is then multiplied by a risk adjusted weighted average cost of capital (WACC) to give a return on the assets employed. This is broadly equivalent to a net margin as defined in our main report (ie. revenues less the cost of sales and operating costs).3

The table below summarises each of these elements and Appendix A discusses them in more detail.

3 It is equivalent to an EBITDA margin, but given the difference between this margin and an EBIT margin is typically modest for these types of businesses and, given the indicative nature of this analysis, any differences are unlikely to be particularly material.




November 2006



Table 2: MMNE capital invested

Investment

Reference to

appendix

Low

Benchmark

value

$m

High

Benchmark

value

$m

Working capital (current assets less liabilities) A1 24.3 32.8 Prudential capital (investment with NEMMCO) A2 24.1 32.6 Fixed assets (IT systems and other) A3 18.1 18.1

Total investment 66.5 83.5 WACC B8 11.6% 16.4%

7.7 13.7 Less credit for cash investment of prudential capital at 5.75% A2 (1.4) (1.9) Less credit for cash investment of seasonal capital at 5.75% A1 (0.2)

Total net return on capital employed 6.3 11.6 Revenue assumed 250.0 250.0 Total required return as a percentage of revenue 2.5% 4.6%

It should be noted that the margins presented here only represent a return on the tangible assets identified. It does not include a return on any intangible asset of a retailer, as might be reflected in its customer base. Therefore the margin estimated under this calculation is likely to be significantly less than one based on benchmark margins. The way in which the market values retail businesses demonstrates this, as section 4 of the main report describes in some detail.

We have been supplied with data that suggests the customer acquisition costs are in the order of $150 per customer contracted. If this value was applied to a customer base of 250,000, the MMNE retailer suggested in this paper would require an investment of $37.5 million to establish that customer base.4 If the WACC is applied to this investment as well, the required margins would increase as follows.

4 Energy Australia provided indicative customer acquisition costs and we have not independently verified these. We are unaware of detailed market information on these costs for a new entrant retailer acquiring customers largely from existing franchised retailers. These costs are likely to vary across customer types and marketing channels and be commercially sensitive. In the absence of any other substantive information, we have applied the Energy Australia data in the calculations presented in this addendum.




November 2006



Table 3: MMNE capital invested – tangible and intangible assets

Investment Low

Benchmark value

High Benchmark

value

Customer direct marketing investment $m 37.5 37.5 WACC 11.6% 16.4%

Return required in non-physical assets $m 4.3 6.1 Total net return non-physical assets from table above (Table 1) $m 6.3 11.6

Total return on all assets $m 10.6 17.4 Net margin based on revenue assumed % 4.3% 7.1%

3.2 Conclusion Based on the assumed size of the MMNE retailer with revenue in the order of $250m per annum, and a customer acquisition cost of approximately $150 per customer, we have calculated a margin required of between 4.3% and 7.1%. This is broadly consistent with the margins represented in the main report, after allowing for the degree of subjectivity involved in undertaking a cost build-up analysis with current market information, as this addendum illustrates. The results should therefore be considered indicative.




November 2006



A Estimating capital employed

A.1 Investment in working capital We have developed a calculation of net working capital by examining the timing of receipts and payments facing the MMNE retailer. The table below describes the timing for receipts and payments used in the determination of a benchmark for net working capital.

A.1.1 Quarterly billed debtor days We understand that standard credit offered is in the order of 14 days however we understand that the average credit taken might be in the order of 18 days. This is consistent with our understanding of the experience of energy retailers as many customers pay on the due date rather than before, and a large number require a reminder notice to effect payment.

We add 45 days of accrued revenue (50% of the billing period) to the outstanding debtor days for quarterly read customers to represent the unbilled income.

Average days outstanding is therefore 45 days plus 18 days for MMNE (63 days).

A.1.2 Monthly billed debtor days We have assumed a benchmark of 14 business days payment terms for business customers. This is comparable to the experience in Victoria where deemed and standing offers to businesses allow credit of 12 days5 which is likely to be extended by a day or two to affect the transfer of payment through the mail.6 We assume that most business customers (who comprise the majority of monthly billed customers) will pay within this timeframe. Combining this with accrued revenue of 15 days (50% of the billing period); we apply an average outstanding period of 29 days for monthly-billed inflows. We note, however, that due to the assumptions made on the volume of monthly customers, assumptions in regard to their payment practices are not material.

A.1.3 Outstanding days for working capital Table A1 below summarises the outstanding days for inflows and outflows for the calculation of working capital for a franchise retailer.

5 http://www.originenergy.com.au/business/files/deemed_contract.pdf, the 12 business day period excludes four non-working days, and hence businesses have 16 days to pay their electricity bill. 6 We believe this assumption is reasonable although other forms of payment may not suffer from such a delay (eg. direct debit, telephone banking), but sometimes might. In any case, the assumption is not material.




November 2006



Table A1: Summary of assumptions of debtor and creditor outstanding days

Outflow/inflow Parameter Description Days

Outstanding

Outflows Energy payments Annual expense, settled through NEMMCO and market contracts 30

Network payments Annual DUoS and TUoS costs to network businesses under contract 30

NEMMCO payments NEM fees and ancillary services 28

Operating costs Salaries and wages and other costs, assumed to be paid twice a month 14

Inflows Monthly bills Annual revenue 29 Quarterly bills Annual revenue 63

We have applied 30 days outstanding for outflows for monthly expenses (payments to the network operators and NEMMCO), 28 days for wholesale market settlement, 14 days for the payment of salaries7 and other operating costs.

Based on benchmark data for the above inflows and outflows, the return on net working capital is defined as:

capitalworkingininvestmentforAllowanceWACCDays

Inflows

Days

Outflowsn

i

n

i =×⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

⎟⎠⎞

⎜⎝⎛

−⎟⎠⎞

⎜⎝⎛

∑∑==

365365

11

Where “Days” is represented by the amount of days the transaction remains outstanding until it is converted to a cash payment or a cash receipt.

The expression in parentheses is the net working capital for a franchise retailer.

Table A2 below summarises the benchmark value range of working capital requirements.

7 Based on the total wage bill for corporate and administration, trading, regulatory, settlements, customer communications, pricing and risk functions and the call centre, and other operating costs.




November 2006



Table A2: Summary of benchmark value of working capital requirements

Asset class Benchmark value

$m

Receivables - Monthly billed customers 1.3 Receivables - Quarterly billed customers 40.2 Payables - Energy (7.2) Payables – Network payments (8.6) Payables – NEM other (0.6) Payables – Operating costs (0.8)

Total (low side) 24.3 Increase for seasonal volatility 35%

Total (high side) 32.8

When calculating the working capital requirements for an entity, we understand that the above calculation is representative of the low end of the scale of requirements for a year, and will not necessarily reflect the seasonal fluctuations that are likely to occur in the industry. Electricity consumption is variable throughout a year and varies with the weather, customer lifestyle, economic influences and consumption patterns driven by installed appliances. In particular, consumption varies in a reasonably consistent way across a ‘typical’ year. Therefore we have allowed for a seasonal factor which increases the value of the required working capital to cater for seasonal fluctuations during the year.

We have been provided with some analysis prepared by EnergyAustralia which suggests that the seasonal fluctuation is in the order of 35% above the flat average for each year, and have allowed for this additional amount in the calculation of the net working capital in order to adequately finance the peak capital requirements.8 When excess capital is not required however, it is assumed that this will be invested on the market at a short term deposit rate.9

It is assumed in this calculation that the peak capital is required for 50% of the year, thereby allowing for the difference between average and peak to be invested for the other half of the year. The value of the credit for the interest earned on the deposit for half of the year is included in Table 1 on page 6.

A similar seasonal variation factor is applied to the Prudential capital requirements.

A.2 NEMMCO prudential guarantee capital A retailer of electricity is required to meet NEMMCO’s prudential requirements in order to trade in energy in the national electricity market. This ensures that the generation sector is paid for energy delivered to the pool, and that the payment is effected in a timely manner by the retailers who purchase energy in the wholesale market on behalf of their customers.

8 KPMG has not independently verified the data behind this assumption. 9 There may well be more efficient ways to manage excess working capital. This assumption is, however, not material.




November 2006



Our calculation for allowed return on prudential capital is based upon:

• an assumed energy purchase for 2005/06;

• multiplying that annual energy expense by:

- NEMMCO’s required days of cover over the number of days in a year (42/365); and

- a volatility factor for the NSW region of the NEM (2.4);10 and

- a seasonal variation rate of an additional 35% for the high side of the capital required to reflect the seasonal fluctuations in operations as discussed in the debtors and creditors section above.

• multiplying this prudential capital by WACC, net of income of 5.74% 11 assumed to be earned by idle capital. This is based on the assumption that an efficient retailer would not actually lodge the full prudential capital requirement (and incur the full cost of capital on the investment lodged). Rather, it would provide a bank guarantee but would have the cash necessary to support the guarantee at hand, on short-term deposit.

A.3 Fixed assets We consider that the MMNE business will require the following assets to manage its business and provide the necessary infrastructure to conduct its activities:

• a customer information system to managing customer data, perform billing and manage the information transfer in the market;

• telecommunications systems for voice and data transfer and B2B transactions;

• trading systems for managing energy purchases; and

• office fit-outs.

Table A3 summarises our estimate of the MMNE’s fixed asset requirements.

10 Volatility factors will vary between states reflecting the various states load factor and other market conditions. Our research into the NEMMCO web site suggests a volatility factor of 2.4 for NSW. 11 5.74% is used as the nominal risk free rate in the WACC calculations in Appendix B.




November 2006



Table A3: MMNE fixed assets

Asset class Benchmark value

$m

Fixed CIS/IT costs 17.0 Telecommunications assets for contact centre 0.5 Trading systems (eg. Openlink) 0.5 Other assets (office fit out etc) 0.1

Total 18.1




November 2006



B Estimating the cost of capital

B.1 Introduction This appendix sets out an assessment of a reasonable indicative rate of return or weighted average cost of capital (“WACC”) for a MMNE for the purpose of providing an input to the cost model. It is indicative only as a thorough assessment of the cost of capital would require a more detailed analysis than allowed for here.

B.2 WACC definition The general basis for calculating a Weighted Average Cost of Capital (“WACC”) expressed in nominal pre-tax terms as follows:

Pre-tax WACC = ke / {1-t*(1-γ)} * E/V + kd*D/V

Where:

ke represents the after-tax cost of equity estimated based on the Capital Asset Pricing Model (“CAPM”)

kd represents the pre-tax cost of debt

t represents the statutory corporate tax rate

γ known as “gamma”, represents the value attributed by investors to each dollar of franking credit generated by the company

E/V represents the proportion of total capital funded by equity

D/V represents the proportion of total capital funded by debt

B.3 Capital structure (D/V and E/V) We assume a gearing assumption (D/V) in the range of 0% to 15%. Markets do not readily provide information highly specific to electricity retailers, particularly those of a MMNE (which will be less diversified than other, larger retailers).

Accordingly we have observed the gearing levels of selected food retail businesses listed on the Australian Stock Exchange, as well as selected publicly listed energy businesses with a retail business segment. Food retailers provide a possible proxy because they retail an essential commodity to a broad market. This information is set out in the table below.




November 2006



Table B1: Gearing levels of comparable publicly listed retail businesses

Company12 Debt/ Total Market capitalisation

Coles Myer Limited 21%

Woolworths Limited 11%

Australian Energy Limited 4%

Australian Gas Light 38%

Origin Energy 69%

Source: Bloomberg

We have also considered the likely difference between the operating structure of the above businesses and energy retailers, and conclude that energy retailers should exhibit fairly low levels of financial gearing, particularly a MMNE. We have therefore provided most reliance on the gearing of Australian Energy Limited.13 This is consistent with the analysis in Section 5 of our main report in regard to vertical integration.

B.4 CAPM cost of equity The table below summarises the parameters underlying the estimation of an after-tax cost of equity (ke).

Table B2 CAPM cost of equity estimate

Parameter Description Low High

Rf Nominal risk free rate 5.74% 5.74%

MRP Market risk premium 6.0% 6.0%

βe Equity beta 0.60 1.12

βa Asset beta 0.60 1.0

βd Debt beta Nil Nil

γ Gamma 35% 0%

Ke CAPM cost of equity 9.3% 12.4%

• The risk free rate of return has been estimated by reference to the benchmark 10 year Commonwealth Government bond yield, and as an average return for the 20 days ended 30 June 2006. This measurement approach represents widely accepted practice for estimating

12 The data on debt and betas is taken at 18 August 2006, using where relevant the most recent 48 observations of monthly data. 13 There are reasons to be additionally cautious about the relevance of some of this data (including the different range of activities some undertake and the recent corporate activity in relation to a couple of them).




November 2006



returns for regulated assets in Australia. Our data is sourced from the Reserve Bank of Australia.

• An estimate of 6% for the MRP remains consistent with the majority of the historical evidence on Australia’s MRP.

• We have adopted an asset beta in the range of 0.6 to 1.0 for MMNE Retailer. Based on a gearing level of nil% to 15%, this translates to an equity beta range of 0.6 to 1.12. Our choice of values is based on a number of considerations including:

- the assessed un-geared beta (i.e. asset beta) values of the comparable businesses set out in Table B2 above. These values are set out in the table below.

Table B3: Equity and Asset betas of comparable publicly listed retail businesses

Company Equity beta Asset beta

Coles Myer Limited 0.91 0.79

Woolworths Limited 0.60 0.34

Australian Energy Limited 1.28 1.25

Australian Gas Light 0.57 0.45

Origin Energy 0.49 0.33

Notes:

1 Data sourced from Bloomberg (using monthly data for past 48 months)

2 Asset beta calculation assumes zero debt beta.

3 The difference between the betas for Coles Myer and Woolworths is likely to reflect the fact that department store operations constitute a part of Coles Myer’s business, whereas Woolworths’ operations are predominantly in supermarkets. Department store operations would tend to be more cyclical by nature.

- the operating differences between the comparable businesses and an energy retail business, and how these difference might translate into differences in relative risk. In particular, we have noted that the only “pure” electricity retailer listed on the ASX (Australian Energy Limited) has a higher beta and has since delisted. It, however operated in the small commercial segment of the market;

- the operating environment faced by a MMNE Retailer, including the risks such as energy purchase costs, and how this might affect the risk faced by the business; and

- acknowledgement that it is difficult to obtain a high level of confidence on what constitutes an appropriate beta value given that most beta measurements tend to be inherently volatile over time.




November 2006



B.5 Value of imputation credits (gamma) We have adopted a range of nil to 35% for the value of imputation credits (i.e. gamma). The evidence that we have reviewed in establishing this range includes:

• Recent research on gamma by Cannavan Finn and Gray (2004)14 which suggests that a value of nil is appropriate for large businesses post the introduction of the 45-day holding period rule for trading in imputation credits;

• A recent analysis by KPMG indicates that standard market practice in relation to gamma, as evidenced by the treatment of the parameter in independent expert reports supporting takeover bids in Australia, is to attribute a value of nil for this parameter; and

• Recent research by Hathaway & Officer (2004)15 suggests that using aggregate tax statistics, the value of gamma is approximately 35%. This research updates previous research by the authors in 1999 which suggest a value of 50% for gamma. The authors’ 1999 research was heavily relied upon by regulators.

The above research suggests that a range of nil to 35% is appropriate.

B.6 Corporate tax rate We have adopted the prevailing statutory corporate tax rate of 30%.

B.7 Pre-tax cost of debt Consistent with current widely accepted practice, the pre-tax cost of debt has been estimated by adding a debt margin to the risk free rate of return.

In estimating the debt margin, we considered what would be a reasonable all-in cost of sourcing long term debt for an energy retail business, given the risks faced by the business, the relatively small debt portfolio that such businesses are likely to hold, and the low likelihood that such businesses would seek a credit rating for the purpose of raising debt in domestic capital markets. Under these circumstances, we conclude that the most efficient source of debt for an energy retail business is bank debt. We estimate that a reasonable margin would lie in the range of 200 to 250 basis points. Assuming a risk free rate of 5.74%, this translates to a pre-tax cost of debt in the range of 7.74% to 8.24%.

14 Cannavan, D, Finn F and S. Gray (2004), The value of dividend imputation credits in Australia, Journal of Financial Economics, 15 Hathaway, N and R.R. Officer (Nov 2004), 'The value of imputation tax credits, update 2004', by Capital Research Ltd.




November 2006



B.8 Summary The table below sets out a summary of our current WACC calculation for a MMNE.

Table B4 WACC estimate

2006 assessment

Low High

Risk free rate 5.74% 5.74%

Equity beta 0.6 1.12

MRP 6.0% 6.0%

CAPM cost of equity (Ke) (Table B2) 9.3% 12.4%

Value of imputation credits 35% 0%

Tax rate 30% 30%

Pre-tax Imputation-adjusted cost of equity

Pre-tax cost of debt 7.7% 8.2%

Equity proportion of capital (E/V) 100% 85%

Debt proportion of capital (D/V) 0% 15%

Pre-tax nominal WACC 11.6% 16.4%

We emphasise that the choice of WACC within the ranges indicated is a matter of judgement. That judgement is complicated by the lack of adequate market data on the key parameters relevant to a MMNE.

The assumptions on which the judgment used in this report are based, are set out above. Different assumptions may result in different judgements, estimates of WACC and hence net margins.

6 A P P E N D I X C – M M A R E P O R T O N A L L O W A N C E F O R W H O L E S A L E C O S T S I N R E T A I L T A R I F F S

Draft Report to

EnergyAustralia

Allowance for Wholesale Costs in Retail Tariffs July 2007 to June 2010

21 December 2006

Ref: J1441v1.3

Project Team

Ross Gawler

Antoine Nsair

Chin Ching Soo

Paul Nidras

Patrick Wang Melbourne Office Brisbane Office 242 Ferrars Street GPO Box 2421 South Melbourne Vic 3205 Brisbane Qld 4001 Tel: +61 3 9699 3977 Tel: +61 7 3100 8064 Fax: +61 3 9690 9881 Fax: +61 7 3100 6067 Email: [email protected] Website: www.mmassociates.com.au

Ref: J1441v1.3, 21 December 2006 McLennan Magasanik Associates

mailto:[email protected]

http://www.mmassociates.com.au/

TABLE OF CONTENTS EXECUTIVE SUMMARY ________________________________________________________ 1

1 INTRODUCTION_________________________________________________________ 8

2 COMPONENTS OF RETAIL COST _________________________________________ 9

2.1 Formulation of Cost Objective and Assumptions ________________________ 10

2.2 Financial assumptions _______________________________________________ 13

2.3 Electricity Tariff Equalisation Fund ____________________________________ 13

2.4 The Load Profile ____________________________________________________ 14

2.5 Estimate of Long-Run Marginal Cost___________________________________ 14

2.6 Energy Price Risk ___________________________________________________ 17

2.7 Contract Prices______________________________________________________ 20

2.8 Demand Growth Uncertainty and the use of Swap Options _______________ 24

2.9 Composite of Contract Prices _________________________________________ 25

2.10 Cap Prices__________________________________________________________ 27

2.11 NGAC Prices _______________________________________________________ 27

2.12 Mandatory Renewable Energy Target __________________________________ 28

2.13 NEMMCO Fees _____________________________________________________ 32

2.14 Ancillary Services Costs______________________________________________ 33

3 WHOLESALE COST ANALYSIS___________________________________________ 35

3.1 Long Run Marginal Cost _____________________________________________ 35

3.2 Contract Coverage __________________________________________________ 36

3.3 Wholesale cost ______________________________________________________ 39

3.4 Abatement and other costs ___________________________________________ 43

3.5 Plexos Modelling Results_____________________________________________ 45

4 CONCLUSIONS _________________________________________________________ 48

APPENDIX A RENEWABLE ENERGY CERTIFICATES MODELLING _____________ 50

APPENDIX B OPTIMISING CONTRACT POSITION TO MINIMISE COST _______ 69

APPENDIX C COMBINING VOLATILITY PARAMETERS IN THE PLEXOS PRICE ANALYSIS _____________________________________________________ 77

APPENDIX D NET SYSTEM LOAD PROFILE ___________________________________ 78

Ref: J1441v1.3, 21 December 2006 i McLennan Magasanik Associates

LIST OF TABLES Table 1 Annual Wholesale Cost (Energy Only) (Nominal Dollars) _____________________ 6

Table 2 Annual Summary Of Additional Abatement Cost and NEM Services with Wholesale Market Cost (Nominal Dollars) ______________________________ 7

Table 2-1 ETEF Reduction _______________________________________________________ 13

Table 2-2 Parameters for Long-Run Marginal Cost Pricing ___________________________ 16

Table 2-3 Analysis of Contract Price data - Acquiring Contract Position from August 2006 – updated as at 22 November 2006._____________________________________ 21

Table 2-4 Risk margin factors for variations in forward contract prices ________________ 22

Table 2-5 Premiums to allow for 12.6% purchase of swap options for final swap tranche _ 25

Table 2-6 Forecast $300 Cap Prices________________________________________________ 26

Table 2-7 New Entry Costs for the Purpose of Estimating LRMC of NGACs ____________ 29

Table 2-8 Renewable Energy Certificate Factors ____________________________________ 29

Table 2-9 Calculation of NGAC Price for Future Years (Nominal Price) ________________ 30

Table 2-10 Renewable energy targets ___________________________________________ 30

Table 2-11 Analysis of REC Prices (Nominal Dollars) _______________________________ 32

Table 2-12 NEM Fees ___________________________________________________________ 33

Table 3-1 Analysis of Wholesale Cost _____________________________________________ 40

Table 3-2 Annual Wholesale Cost (Energy Only) ___________________________________ 43

Table 3-3 Quarterly analysis of additional abatement cost and NEM services ___________ 44

Table 3-4 Annual summary of additional abatement cost and NEM services____________ 45

Table 4-1 Summary of Prices ____________________________________________________ 49

LIST OF FIGURES

Figure 1 Regression of Daily Energy _______________________________________________ 3

Figure 2 Trend in Threshold Temperatures _________________________________________ 4

Figure 3 Swap Contract Coverage _________________________________________________ 5

Figure 4 Example of Basis for Optimal Flat Contract Volume (Q1 2006) _________________ 5

Figure 2-1 Uncertainty of Wholesale Cost (After Hedging) ___________________________ 11

Ref: J1441v1.3, 21 December 2006 ii McLennan Magasanik Associates

Figure 2-2 Typical Load and Contract Position _____________________________________ 17

Figure 2-3 Correlation of Weekly Average NSW Pool Prices 1997 – 2006 _______________ 19

Figure 2-4 Conservative Monotonic Price Versus Demand ___________________________ 19

Figure 2-5 History of Contract Price Trends________________________________________ 23

Figure 2-6 Trend in NEM Fees (Nominal Dollars)___________________________________ 33

Figure 2-7 History of Customer Ancillary Service Cost $/MWh ______________________ 34

Figure 3-1 Long-Run Marginal Cost for 50% POE Peak Demand______________________ 35

Figure 3-2 Ratio of Volume to Time Weighted Price_________________________________ 36

Figure 3-3 Load and Swap Volume for January to March 2009________________________ 38

Figure 3-4 Swap Contract Coverage ______________________________________________ 39

Figure 3-5 Quarterly Prices ______________________________________________________ 42

Figure 3-6 Historical and Forecast NSW Pool Prices _________________________________ 46

Figure 3-7 Historical and Forecast NSW Price Duration Curves _______________________ 46

VERSION Version Date Comment Approved Draft 0.1 18 August 2006 Preliminary working draft for information

only to EA – some sections have yet to be completed

Ross Gawler

Draft 0.2 22 August 2006 Added analysis of wholesale energy costs Ross Gawler

Draft 0.3 25 August 2006 Added Executive Summary. Further changes will be required after 28 August meeting with Energy Australia

Ross Gawler

Draft 0.4 29 August 2006 Editorial changes following meeting with EnergyAustralia on 28th August. Corrected LRMC fixed gas cost modelling. (Not issued to EA)

Draft 1.0 31 October 2006 Amended report to discuss Net System Load as the Basis

Ross Gawler

Draft 1.1 22 November 2006 Amended in response to comments from Phil Moody. Updated information on REC spot price.

Ross Gawler

Draft 1.2 7 December 2006 Reviewed comments from Phil Moody Ross Gawler

Draft 1.3 12 December 2006 Updated analysis using 22 November 2006 contract prices and evaluated wholesale cost

Ross Gawler

Ref: J1441v1.3, 21 December 2006 iii McLennan Magasanik Associates

for 2007/08. Draft 1.4 21 December 2006 Amended inconsistencies between text and

table figures.

Ref: J1441v1.3, 21 December 2006 iv McLennan Magasanik Associates

EXECUTIVE SUMMARY

IPART has recently commenced its review of NSW regulated retail electricity tariffs and charges for customers using less than 160 MWh of electricity per annum. As an incumbent retailer, EnergyAustralia has an interest in the outcome of IPART's review and, as such, intends to be an active participant in the review process.

In order to advance its position, EnergyAustralia engaged McLennan Magasanik Associates (MMA) to help formulate a position on an appropriate allowance for energy costs to be considered by IPART when setting regulated retail electricity prices. The wholesale cost components that go into formulating the retail cost were analysed with supporting analysis.

The wholesale cost of energy used to supply retail customers using less than 160 MWh per annum is made up of the following components:

• The wholesale spot market energy cost purchased at the Sydney West Reference node.

• The costs of hedging the risk of this wholesale energy purchase by means of trading electricity derivatives at the NSW reference node, that is, the hedge contract costs.

• Charges for operating in the NEM paid to NEMMCO for management and ancillary services.

• The sum of these prices is multiplied by the marginal transmission loss factor that applies for a particular connection point.

• This price is further multiplied by a distribution loss factor that refers the wholesale cost to a particular customer meter.

• The price of purchasing NSW Greenhouse Gas Abatement Certificates (NGACs) to satisfy the requirements of the NSW Greenhouse Gas Abatement Scheme (GGAS) established by the NSW Government.

• The price for purchasing Renewable Energy Certificates (RECs) to meet the Mandatory Renewable Energy Target (MRET) established by the Commonwealth Government.

The assumptions made in the analysis were those considered prudent for a mass market retailer which has been in business over the last three years in preparation for serving a stable retail portfolio in the period July 2007 to June 2010. Whilst allowance has been made for volume uncertainty, the pricing has been developed for a constant set of customers with growing per customer usage in accordance with trends derived from data provided by EnergyAustralia for the Net System Load Profile plus the Controlled Load, represented symbolically as “NSLP+CLP” or “net system load”1. This profile only applies

1 It should be noted that the “NSLP + CLP” we are referring to are those relating only to the EnergyAustralia

franchise area. The cost implications are different for each franchise area's NSLP & CLP because they may differ according to the customer mix. The (NSLP + CLP) is not simply the mathemtical sum of the two loadshapes, but rather a weighted sum by half-hour that reflects the changing volume over time.

Ref: J1441v1.3, 21 December 2006 McLennan Magasanik Associates 1

to the EnergyAustralia region and may not be applicable for other regions or retailers. The profile was developed by combining the net system load adjusted for the Hydro Aluminium load with the controlled load shape adjusted to match the financial year sales volumes. An exponential time function was applied in each half-hour to match the financial year total energy.

The changing intensity of heating and cooling load has been identified in the historical data.

A number of analyses were conducted as follows:

• The historical relationship between spot price and net system load in 2005/06 was analysed to determine a model of the spot market exposure as determined from the load profile and the correlation with spot price. A separate analysis was conducted for peak and off-peak periods.

• The historical pattern of the 2003 to 2006 net system load in relation to hourly temperatures and day type was analysed to determine a formulation of the daily energy, peak energy and peak demand as a function of heating and cooling degree days with trend over time. This enabled a close fit of the consumption pattern to be developed as shown in Figure 1 for the daily energy. It also enabled the trends in load magnitude to be separated from the growth in heating and cooling load.

• It was found that the heating and cooling loads were both growing and that the threshold temperatures for calculating heating and cooling degree days were slowly trending towards each other as shown in Figure 2. This trend was extrapolated in the analysis of future time periods as shown in the Figure. It is possible that extrapolation of this trend beyond 2010 would be impractical because the heating and cooling thresholds would cross over. The regression analysis indicated much higher rates of change at more than 1.0o C/year shown but that was not credible as a linear trend to be extrapolated. To obtain more sensible results the rate of change was limited to 0.5o C/year. There is scope for further analysis to assess the trends in threshold temperature. An exponential trend to a new steady state temperature might be an alternative hypothesis that would be worth testing in future work.

• The trend in the parameters together with the 2005/06 weather was used to forecast a net system load shape that would be consistent with a pool simulation using the same load shape. Load shapes were developed for 90%, 50% and 10% probability of exceedance (POE) weather by amending the extreme days in the 2005/06 weather year to fit the required temperature profile as published by TransGrid for each of these conditions. By this means we have a net system load shape and a system load shape that are consistent and it is easier to assess volume and price risk for different hedging positions.


Figure 1 Regression of Daily Energy

Daily Energy for Net System Load + Controlled Load 2004/05

15000200002500030000350004000045000500005500060000

1/07

/200

4

15/0

7/20

04

29/0

7/20

04

12/0

8/20

04

26/0

8/20

04

9/09

/200

4

23/0

9/20

04

7/10

/200

4

21/1

0/20

04

4/11

/200

4

18/1

1/20

04

2/12

/200

4

16/1

2/20

04

30/1

2/20

04

13/0

1/20

05

27/0

1/20

05

10/0

2/20

05

24/0

2/20

05

10/0

3/20

05

24/0

3/20

05

7/04

/200

5

21/0

4/20

05

5/05

/200

5

19/0

5/20

05

2/06

/200

5

16/0

6/20

05

30/0

6/20

05

Day

MW

h

FitActual


150002000025000

30000350004000045000

5000055000

1/07

/200

5

15/0

7/20

05

29/0

7/20

05

12/0

8/20

05

26/0

8/20

05

9/09

/200

5

23/0

9/20

05

7/10

/200

5

21/1

0/20

05

4/11

/200

5

18/1

1/20

05

2/12

/200

5

16/1

2/20

05

30/1

2/20

05

13/0

1/20

06

27/0

1/20

06

10/0

2/20

06

24/0

2/20

06

10/0

3/20

06

24/0

3/20

06

7/04

/200

6

21/0

4/20

06

5/05

/200

6

19/0

5/20

06

2/06

/200

6

16/0

6/20

06

30/0

6/20

06Day

MW

h

FitActual


20000

25000

30000

35000

40000

45000

50000

55000

1/07

/200

3

15/0

7/20

03

29/0

7/20

03

12/0

8/20

03

26/0

8/20

03

9/09

/200

3

23/0

9/20

03

7/10

/200

3

21/1

0/20

03

4/11

/200

3

18/1

1/20

03

2/12

/200

3

16/1

2/20

03

30/1

2/20

03

13/0

1/20

04

27/0

1/20

04

10/0

2/20

04

24/0

2/20

04

9/03

/200

4

23/0

3/20

04

6/04

/200

4

20/0

4/20

04

4/05

/200

4

18/0

5/20

04

1/06

/200

4

15/0

6/20

04

29/0

6/20

04

Day

MW

h

FitActual

• It was initially assumed that a prudent retailer would acquire a flat and peak swap position over a period of three years leading up to the financial year of trading. However, in view of the recent announcement of the phasing out of ETEF, the costing analysis assumed that contract positions are taken up from 2006/07 over less than three years where necessary based upon lead time.

• It was assumed that a $300 cap would be purchased to cover the gap between the peak swap capacity and the 10% POE peak demand. This is efficient because the average volume of energy in this capacity band is very small.


Figure 2 Trend in Threshold Temperatures

Trend in Threshold Temperatures

15

16

17

18

19

20

21

22

23

24

2003 2004 2005 2006 2007 2008 2009 2010

July

Degr

ess

Cels

ius Peak Demand Summer

Peak Energy SummerDaily Energy SummerPeak Demand WinterDaily Energy WinterPeak Energy Winter

Actual Trend Extrapolated Forecast

• Contract prices for flat and peak swaps and flat caps were obtained from d-cypha and NGeS sources for the years from 2007 to 2010 on 22 November 2006. These contract prices were used to estimate future financial year contract prices having regard to the historical volatility of contract prices. A final contract purchase price was estimated by allowing for 90% probability that the actual contract price would be less than the chosen value. The contract prices will need to be reviewed closer to when the final wholesale price is determined.

• Using the 2005/06 system load profile, a PLEXOS simulation was applied to provide estimates of future peak and off-peak prices.

• These spot prices were applied to estimate optimal contract volumes that would minimise the wholesale energy cost including spot and contract costs at the level which has a 90% probability of exceeding the actual future cost. It was found that the optimal flat contract level was within 10% of the off-peak average load and the optimal peak period contract volume was also within 10% of the average peak period load as shown in Figure 3. It may be noted that a higher proportion of the average load may be contracted in Q2 when prices are less volatile than at other times of the year.

• In the later periods of the study, from mid 2009, the mismatch between spot and contract prices was such that the optimal contract volume was infinite based upon the analysis. This was because the impact of extreme events was excluded. The contract price was then adjusted so that an optimum volume was obtained with the upside and down side exposure versus volume about equal. An example of this outcome from the historical load/price analysis is shown in Figure 4.


Figure 3 Swap Contract Coverage

Ratio of Swap Contract Volume to Average Load

94%

96%

98%

100%

102%

104%

106%

108%

110%

112%

Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2

2007 2008 2009 2010

Quarter

Off-Peak PeriodPeak Period

Figure 4 Example of Basis for Optimal Flat Contract Volume (Q1 2006)

Cost for Flat Contract for Q1 2006

$70.00

$70.50

$71.00

$71.50

$72.00

$72.50

$73.00

200 300 400 500 600 700 800 900 1000 1100

Flat Contract Volume (MW)

$M

Average off-peak load

• The wholesale purchase costs using market based hedges was then estimated for the preferred contract volume by summing the spot and contract costs and then dividing by the energy volume under the load shape. The resulting hedge costs varied from


$55.40/MWh in 2007/08 to $57.06/MWh in 2009/10 as shown in Table 1 in dollars of the year.

Table 1 Annual Wholesale Cost (Energy Only) (Nominal Dollars)

Financial Year 2008 2009 2010Ending June Units

Wholesale Cost $/MWh $55.40 $56.92 $57.06

LRMC Estimate $/MWh $56.53 $60.09 $62.03

ETEF Share 100% 77% 37%

Note: Reflects costs at NSW RRN, excludes distribution and transmission losses.

• A long-run marginal cost (LRMC) model was also developed for comparison purposes using black coal fired plant at Mt Piper, combined cycle gas fired plant at a coastal location, an open cycle gas turbine and a demand side response for the extreme peaks. The cost parameters included uncertainty for each cost element. A random simulation with 1,000 samples was executed to obtain the expected cost and its standard deviation. The samples were very close to a Normal distribution. The mean and standard deviation were used to estimate a value for the LRMC that would have an 80% probability of exceeding the outcome.

The prices for the LRMC were determined annually and the wholesale cost from trading in the energy market was assessed quarterly. The results are summarised in Table 1.

To these costs were added the costs of:

• RECs for the MRET scheme;

• NGACs for the NSW GGAS based upon an estimated ratio of NGACs per wholesale MWh;

• NEM fees charged by NEMMCO; and

• NEM ancillary services attributed to customers.

These items add a further $5 to $6/MWh and give the total energy costs shown in Table 2 for the wholesale market cost. It ranges between $60.80/MWh in 2007/08 to $63.25/MWh in 2009/10 rounded to the nearest 5c. The Table shows the assessed costs and proportions of emission abatement purchases per MWh at the wholesale level.

The analysis has demonstrated the changes in the patterns of loading and the uncertainty in the key cost components. It will be important for IPART in setting the regulated tariff to ensure that the trading range allows for these uncertainties and minimises the risk of market failure by not setting the regulated tariff too low. Each prudent retailer would need to operate on the same basis and would be prepared to pay a premium to ensure that its wholesale cost risk is properly managed relative to its retail obligations. Since each


retailer would face the same risks each would be expected to pay similar premium levels to manage the wholesale price risk.

Table 2 Annual Summary Of Additional Abatement Cost and NEM Services with Wholesale Market Cost (Nominal Dollars)

Financial Year 2008 2009 2010Ending JuneREC Price $/MWh $30.87 $34.28 $37.35Renewable Power Percentage 2.86% 3.33% 3.82%REC Component $/MWh $0.88 $1.14 $1.43

NGAC Price $/NGAC $14.41 $15.46 $15.09NEC Ratio NGAC/MWh 0.268 0.266 0.275NGAC Component $/MWh $3.86 $4.12 $4.16

NEM Fee Levy $/MWh $0.35 $0.33 $0.32

Ancillary Services Charge $/MWh $0.30 $0.30 $0.30

Additional Cost $/MWh $5.39 $5.89 $6.21


Total Wholesale Supply Cost $/MWh $60.79 $62.81 $63.26

The cost levels are subject to a degree of uncertainty related to the supply/demand trends in the NEM and the cost factors affecting long-run marginal costs. These cost components may need to be reviewed just before the final pricing decision is made.


1 INTRODUCTION

IPART has recently commenced its review of NSW regulated retail electricity tariffs and charges for customers using less than 160 MWh of electricity per annum. As an incumbent retailer, EnergyAustralia has an interest in the outcome of IPART's review and, as such, intends to be an active participant in the review process.

In order to advance its position, EnergyAustralia engaged McLennan Magasanik Associates (MMA) to help formulate a position on an appropriate allowance for energy costs to be considered by IPART when setting regulated retail electricity prices. This report discusses the wholesale cost components that go into formulating the retail cost with supporting analysis.

A previous report entitled “Market Assumptions for Estimating Wholesale Costs” has detailed the assumptions concerning the National Electricity Market (NEM) that have been used for marketing modelling to support the price risk analysis. This report summarises the results of the analysis that have been undertaken to determine the basis for a wholesale cost.

The cost analysis has been undertaken using historical data for the Net System Load Profile plus the Controlled Load Profile, represented symbolically as “NSLP+CLP” or “net system load”. In this report reference to the “net system load” in most cases includes the addition of the controlled load except where the addition is explicitly stated.

Ref: J1441v1.3, 21 December 2006 8 McLennan Magasanik Associates

2 COMPONENTS OF RETAIL COST

The wholesale cost of energy used to supply retail customers using less than 160 MWh per annum is made up of the following components:

• The wholesale spot market energy cost purchased at the Sydney West Reference node.

• The costs of hedging the risk of this wholesale energy purchase by means of trading electricity derivatives at the NSW reference node, that is the hedge contract costs.

• Charges for operating in the NEM paid to NEMMCO for management and ancillary services.

• The sum of these prices is multiplied the marginal transmission loss factor that applies for a particular connection point.

• This price is further multiplied by a distribution loss factor that refers the wholesale cost to a particular customer meter.

• The price of purchasing NSW Greenhouse Gas Abatement Certificates (NGACs) to satisfy the requirements of the NSW Greenhouse Gas Abatement Scheme (GGAS) established by the NSW Government.

• The price for purchasing Renewable Energy Certificates (RECs) to meet the Mandatory Renewable Energy Target (MRET) established by the Commonwealth Government.

The purpose of this project is to estimate the costs of these service elements over the period from July 2007 to June 2010. The basis for the cost estimate would be that of a new entrant retailer which implies that any existing committed costs incurred by incumbent retailers are irrelevant unless they would be incurred by a prudent retailer.

An electricity retailer operating in a competitive market would operate on quite small net margins of some 2%to 5% of gross sales depending on the risk of supplying the groups of customers. Since spot prices in wholesale electricity markets averaged over a year can vary by plus or minus 30%, it is necessary for a prudent retailer to use financial instruments to hedge the spot price risk so that it can offer its retail customers a fixed price. The purchasing of emission abatement products will also require forward purchasing and spot trading or banking to manage retail obligations. Since sales volumes are also variable according to weather and customer churn, there is also a substantial volume risk when the contract volume is mismatched with the customer load.

The electricity hedge contract market operates up to 5 years in advance. Normally, hedge contracts are entered into for 2 to 3 years. It is unusual for retailers to contract much volume beyond a 5 year period because it is rare for their customers to make contractual commitments for more than 2 to 3 years at a time. Ideally a prudent retailer would build up its contract portfolio over a period of time and would purchase on a longer time frame if low cost resources are offered for purchase. It would seek to add to its portfolio when

Ref: J1441v1.3_PM, 21 December 2006 9 McLennan Magasanik Associates

contract prices are lower on a speculative basis. It may reduce its contract exposure when contract prices rise, expecting with some confidence to be able to buy back its contract cover later at a lower price. It would also need to reduce its contract cover if it experiences loss of customer load to minimise the risk of having to make unfavourable difference payments to generators during low price periods.

Some trading profits can be made by speculating on contract positions but there is also the risk of losses if market price trends diverge from expectations. In formulating the cost of retail service, it is not appropriate to include any savings from speculative trading because such activities are not directly related to retail supply. They have their own structure of required skills, costs and benefits quite separate from that required to arrange retail supply. Therefore, we model a time based contract purchasing strategy to gain the advantage of exposure to average wholesale prices to match the market.

There remains the possibility that a cost estimated in this way could under-state or over-state the costs of a retailer entering the market in 2006/07 without having committed to any contracted resources in prior years. This is a risk that IPART will need to manage as the regulator. The best that can be done at this stage is to present the available data and a rational argument to provide the basis for an assessment. There is no guarantee that the results will reflect the costs incurred by any particular retailer.

The costs assessed here were based on the retail profile data provided by EnergyAustralia for its net system load profile. MMA does not have access to data for any other party.

2.1 Formulation of Cost Objective and Assumptions

2.1.1 Objective

In this study we estimate a static view of wholesale cost based upon the following objectives:

1. The retailer will endeavour to achieve a high level of hedging for its customer load using standard traded instruments: flat swaps, peak period swaps, and caps with $300/MWh strike price. It will not be able to entirely eliminate risk because it cannot cover all of its load exactly in all settlement periods by purchasing a load following contract as the cost of these contracts is high and supply is limited.

2. The retailer will attempt to minimise its wholesale cost at a 90% probability of coverage of its wholesale energy cost. This means that it focuses on a cost measured on a probability distribution at the 90% probability level as illustrated in Figure 2-1 and it contracts accordingly. This means that load coverage might be adjusted according to hedge costs from time to time if it could reduce the “risk managed cost”.


Figure 2-1 Uncertainty of Wholesale Cost (After Hedging)

The optimal contracting level would also depend on the extent to which a large retailer can move the spot and contract market price by changing the volume contracted. As discussed below in section B.3 of Appendix B , we have not attempted to estimate the importance of this effect because it depends on the size of the contracting business relative to the market and the prevailing market conditions in ways that are difficult to quantify without quite specific assumptions and much supporting data which are not readily available.

$/MWh

Probability Density

OBJECTIVE: Minimise the 90% confident cost level 90% of the

area

2.1.2 Energy Market Assumptions

In order to achieve this objective, the following assumptions are made as representing a prudent approach to risk management and costing in the energy market:

1. EnergyAustralia currently has coverage under the Electricity Tariff Equalisation Fund (ETEF) for its regulated load. It is assumed that the energy cost allowance in the next retail price determination will be based on market based hedge costs and that the strike price of the ETEF in future will continue to be set equivalent to the energy cost allowance of the determination to foster competitive neutrality across retailers.

2. LRMC in this study was based upon estimated costs for new black coal fired plant at Mt. Piper, gas fired combined cycle plant such as Tallawarra, open cycle gas fired plant and demand side response for duties less than 4 hours per year.

3. A retailer will build up its hedge position over a period of three years leading up to the time of supplying the customer. This means that the cost of a given hedge will reflect


the average price over the three years prior to the service period1. In formulating the contract price for EnergyAustralia we have recognised that the volume which replaces ETEF would not have been able to be contracted until after March 2006 when the change in ETEF arrangements was announced.

4. The retailer will purchase swap options to cover that part of its forecast load which it considers to be less than 80% confident of retaining. This requires a measure of peak and off-peak sales volume uncertainty.

5. The retailer will sell out of cap positions without loss of value if customer volume is lost. This assumes a zero sum game in that loss of a customer is associated with gain of the customer by another retailer who will wish to purchase the cap cover at the market price.

6. Any optimisation of the retailers’ hedge position by timing the acquisition of contract positions and buying low and selling high during this run-up period is regarded as a trading activity and not a wholesale cost. Such trading requires sophisticated technology and staff that are not specifically required to hedge a retail portfolio. Accordingly, the costs and benefits of such trading are not included in the retail cost.

7. The spot market price for the purposes of estimating residual volatility of cost was estimated from a range of methodologies to reflect the uncertainty of developments in the NEM. The prospective methods include:

• Using historical price traces from the previous three years sorted according to daily energy and day of the week in each month. This is of limited value because there are only three samples of possible price outcomes.

• Using PLEXOS simulations for 10%, 50% and 90% probability of exceedance (POE) peak loads with the system load shape matched to the retail load shape. This is of limited value because it is difficult to replicate the dynamics of bidding in a conventional stochastic model without modelling agent behaviour in detail.

• Developing a stochastic model of price volatility in peak and off-peak periods from price outcomes over a longer period and scaling the parameters to match expected future supply/demand conditions. This is limited in value because it is difficult to capture possible changes in contracting and bidding patterns.

4. We developed a composite approach which involved using the PLEXOS model to estimate the future hourly patterns of spot prices and developing a combination of the historical relationship between price volatility and net system load as described in

1 It has been noted by EnergyAustralia that the three year time frame would be relevant to a private retailer which did not

have access to ETEF. For EnergyAustralia, it has only been advised about the changes to ETEF in March 2006 so it would not have been able to contract its position during 2004/05 and 2005/06 when forward contract prices were lower. It is not clear how IPART is going to resolve this differing exposure to the wholesale market by Government owned and private retailers.


Appendix B and the observed relationship from the PLEXOS results. The results of this analysis are described in Appendix C .

2.2 Financial assumptions The other assumptions which formulated the basis of this study, are:

• 3% pa CPI escalation from June 2006 throughout the period

• Real interest rates of 5% pa applied to holding assets such as RECs and NGACs. For the 3% CPI assumption this is equivalent to a nominal interest rate of 8.15%.

2.3 Electricity Tariff Equalisation Fund The role of the ETEF in managing the wholesale cost of electricity for the regulated segment of the retail market is to be reduced in accordance with the schedule in the IPART Issues Paper as shown in Table 2-1.

Table 2-1 ETEF Reduction

Inclusive Dates Percentage of NSW regulated retail load supported by ETEF

Prior to September 2008 100%

From September 2008 to February 2009 80%

From March 2009 to August 2009 60%

From September 2009 to February 2010 40%

From March 2010 to June 2010 20%

The IPART Issues paper states as follows:

“As ETEF is phased out, standard retailers will be responsible for managing the exposure to pool price risk for an increasing portion of the load associated with regulated retail customers. The Tribunal’s terms of reference for this review require it to consider a number of matters related to the phasing out of the ETEF in making its determination on retail electricity tariffs. The matters include:

• An allowance for electricity purchase costs based on an assessment of the long run marginal cost of electricity generation from a portfolio of new entrant generation to supply the load profile of customers remaining on regulated retail tariffs – that is, not based on forecast market based prices. (This issue is discussed in detail in section 4.1 [of the Issues Paper].)

• The hedging, risk management and transactions costs that retailers will face in the absence of the ETEF.

• The forecasting risks that retailers will face in the absence of the ETEF.”


In section 4.1 of the Issues Paper, the basis for the current ETEF price is stated as follows:

“The current determination incorporates a LRMC based on new entry generation of $47 per MWh (in 2004/05 dollars). The Tribunal adopted this estimate of the LRMC, based on the medium cost scenario presented by Intelligent Energy Services (IES), one of its consultants for the 2004 review. The LRMC includes an allowance of $0.10 for generator National Electricity Market (NEM) fees, as these fees are unavoidable costs incurred by generators in the process of selling electricity in the NEM.”

Supporting the principle of competitive neutrality across retailers, the strike price of the ETEF should be based on wholesale market hedge costs, which in turn should underpin the energy cost allowance in the next retail price determination.

2.4 The Load Profile Whilst allowance has been made for volume uncertainty, the pricing has been developed for a constant set of customers with growing per customer usage in accordance with trends derived from data provided by EnergyAustralia for the Energy Australia Net System Load Profile [NSLP] and Controlled Load Profile [CLP], represented symbolically as “NSLP+CLP” and referred to in this report as the “net system load”2. This profile only applies to the EnergyAustralia region and may not be applicable for other regions or retailers. The profile was developed with reference to the historical NSLP adjusted for the Hydro Aluminium load with the controlled load shape adjusted to match the financial year sales volumes. An exponential time function was applied in each half-hour to match the financial year total energy. This profile represents the average ‘small’ contestable customer base a MMNE retailer would be exposed to and is also consistent with the regulated load shape.

2.5 Estimate of Long-Run Marginal Cost In this study we have assessed an equivalent LRMC for the coming three years based on the following assumptions in June 2006 dollars:

Base load power is supplied by a new coal fired power station at Mt Piper. A cost of $1,500/kW for the plant and $1.40/GJ for the fuel was allowed. The ACIL Tasman report of February 2005 indicated an average fuel cost to the Western Power stations of $1.46/GJ. We would expect that supply to a new power station would provide some negotiation strength and have therefore allowed a coal cost of $1.40/GJ in June 2006 dollars with a standard deviation of uncertainty at 10%. Coal costs could be higher or lower depending on the extent of competition for new coal contracts.

2 It should be noted that the “NSLP + CLP” we are referring to are those relating only to the EnergyAustralia

franchise area. The cost implications are different for each franchise area's NSLP & CLP because they may differ according to the customer mix. The (NSLP + CLP) is not simply the mathemtical sum of the two loadshapes, but rather a weighted sum by half-hour that reflects the changing volume over time.


Intermediate load power is supplied by a new combined cycle gas fired plant at a coastal location. The plant cost was assumed to be $1,000/kW and the fuel to be delivered at $3.90/GJ with fuel cost rising at CPI-1%3.

Peak load is supplied by an open cycle gas turbine located at Munmorah with a capital cost of $700/kW. A $3/GJ is added to the gas cost to reflect a low duty supply. A small part of cost (50c/GJ) is treated as a variable cost and the remaining $2.50/GJ is treated as a fixed cost at 8% capacity factor. This capacity factor corresponds to the break-even capacity factor with combined cycle plant.

Extreme peak demand is supplied by demand side response. The key parameters and their uncertainties are shown in Table 2-2. The parameters are based on a November 2005 presentation by Energy Response4.

MMA’s approach to estimating LRMC includes an estimated allowance for uncertainties in costs and then to derive a risk adjusted LRMC at an 80% one-sided confidence interval5. We have not attempted to quantify all the possible risk factors on costs, such as exchange rates and the global supply and demand of power generation equipment. We have proposed some indicative factors for uncertainty based on industry experience.

All the cost components are assumed to be fully correlated and the heat rate is negatively correlated with the costs. That is, the heat rates are lower if the prices are higher. For this particular set of assumptions, the expected LRMC for the 2005/06 net system load profile for EnergyAustralia was estimated to average $52.60/MWh. The LRMC which has an 80% probability of coverage of the range of uncertainty was $57.70/MWh with a risk premium of $5.10/MWh in June 2006 dollars. This assessment includes allowance for the marginal revenue earned from NGACs at $11/NGAC at 0.95t/MWh pool coefficient. Chapter 3 discusses the results for the forecast years.

This value is higher than the $47/MWh previously assessed by Intelligent Energy Systems (IES) in December 2004 ($49.50 in June 2006 dollars). The factors which would have contributed to the increase would be escalation in costs (the cost components used by IES were not readily available to MMA) and the lower load factor due to the increase in air-conditioning demand.

3 This CPI-1% price change represents the likely trend in new contract prices from year to year over this period. From 1985

until 2000, gas prices for new contract stayed about $3/GJ which was about CPI-3%. Since 2000, gas prices have risen more in line with CPI but because of increasing competition in the gas market. MMA is projecting a slight fall in real gas prices over the period to 2010. This is in line with the long-term trend in wholesale electricity prices.

4 The presentation maybe found at http://www.sustainability.vic.gov.au/resources/documents/C3_Ross_Fraser.pdf 5 The LRMC has a 90% confidence of coverage of possible outcomes based upon an estimate of the standard deviation of the

cost and using the Normal distribution approximation.


Table 2-2 Parameters for Long-Run Marginal Cost Pricing

Parameter Units (June 2006 Dollars)

New coal fired technology

Combined Cycle Plant

Open Cycle Plant

Demand Side Response

CPI Trend

Uncertainty Measure Basis Comments

NGAC Value $/t NGAC $11Pool Coefficient t/MWh 0.95

Overnight Capital Cost$/kW

$1,500 $1,000 $700 $100-1%

10%Standard Deviation of Cost For power plants

50%Uniform Range for DSR For demand side response

Economic Life Years Years 30 25 25 10 5 Range of Years Uniform DistrbutionInterest During Construction

% real

9.29% 7.67% 4.52% 0.00% 10%

As a Uniform Range

Based upon 5% real interest rate

Marginal Transmission Loss Factor

0.9756 0.9887 0.9887 1.0200 0.02 Uniform Range for gas and DSR

CCGT at Munmorah, Coal at Mt Piper

Fixed Operating Cost$/kW/year

$35 $20 $12 $9 0% 20%Standard Deviation of Cost

Variable Operating Cost$/MWh

$2 $3 $5 $2,000 0% 20%Standard Deviation of Cost

Variable Fuel Cost

$/GJ HHV

$1.40 $3.90 $4.40 $0.00 -1% 10%

Standard Deviation of CostBased on MMA-Gas Model for gas. $0.50/GJ premium for peaking duty

Fixed Fuel Cost

$/GJ HHV

$2.50 -1% 10%8% capacity factor for $2.50/GJ fixed cost

peaking gasHeat Rate

GJ/MWh9.2 7.3 11 0 2%

Standard Deviation of ValueFuel CO2 Rate

kg/GJ90 56 56 0 2%

Standard Deviation of ValuePlant Availabiity 90% 92% 94% 95% 2% Range of ValuesEfficient Dispatch Capacity Factor without abatement %

90% 42.5% 8.5% 0.480% 2%Range of Values for Coal only CCGT by Calculation

versus Capability Coal Plant CCGT OCGTEfficient Dispatch Capacity Factor with abatement at $11t/CO2

%

90% 59.7% 7.7% 0.482% 2%

Range of Values for Coal only CCGT by CalculationWeighted Average Cost of Capital (Real Pre-Tax) Real Pre-Tax

7.80% 7.80% 8.50% 8.50% 0.5%

Absolute Range Uniform DistrbutionTotal Fixed Cost to Pool

$/kW/year

$202.6 $131.0 $110.6 $25.0Fixed Cost DSR based on Energy Response Nov-05

Emission Rate t/MWh t/MWh 0.828 0.4088 0.616 0Total Fixed Cost to Pool $/kW/year $202.6 $131.0 $110.6 $25.0Total Variable Cost to Pool $/MWh $14.88 $31.50 $53.44 $2,000.00Total Variable Cost Less Emission Abatament $/MWh

$13.54 $25.54 $49.77 $1,989.55


2.6 Energy Price Risk The difficulty in estimating energy price risk stems from the following factors:

• The retailer is not fully able to match hedge volume to wholesale purchase volume reflecting customer demand. This is illustrated in Figure 2-2 which shows a load profile and a typical swap contract position over a daily load profile.

Figure 2-2 Typical Load and Contract Position

Load Volume

Time

Swap Contract Position

Load Profile

Exposed to low prices if demand and price are low

Exposed to high prices if demand and price are high

The hedge is referred to as a swap because the two parties swap payments of the fixed contract price for the variable Spot Price. This is also described as a “contract for differences” because the net payment is the fixed price less the spot price paid by the contract purchaser to the contract seller. The net Cost to the retailer is as shown in Equation (1):

Cost = Contract Volume* (Contract Price – Spot Price) + Load * Spot Price (1)

This may be rearranged to better show the importance of the contract volume and load as shown in Equation (2):

Cost = Contract Volume* Contract Price + Spot Price * (Load – Contract Volume) (2)

If the Contract volume is equal to the Load then the Spot Price coefficient in equation (2) is zero and the total cost is fixed at:

Intended Fixed Cost = Contract Volume* Contract Price (3)

This is an ideal position for a retailer from a risk viewpoint as long as the retailer has not paid too much for the contract. If it did, its cost would be fixed but it would be at risk of losing retail volume which would undermine the hedged position over time as


retail customers switch to lower cost providers. If it competes by lowering its retail price, it would suffer financial loss and threaten the viability of the business.

If Load < Contract Volume, as might occur during off-peak periods then the Cost would be reduced as long as the Spot Price does not go negative as can sometimes occur if too much generating plant is committed. If the Spot Price goes very high then the retailer would receive a windfall income.

If Load > Contract Volume, as might occur at peak times when weather is more extreme that unusual or otherwise forecast, then the Cost becomes exposed to positive price spikes with high magnitude and duration as long as the Load continues to exceed the Contract Volume. If Spot Prices are low then the volume imbalance favours the retailer as the Cost decreases with Spot Price

The high price risk due to insufficient volume can be mitigated economically by purchasing one-way hedges (also described as “caps” because they provide a cap on the exposure to high spot prices). These hedges pay a difference payment whenever the spot price exceeds the strike price of the cap. Strike prices are typically $200/MWh or $300/MWh for this purpose. This is a financial equivalent to buying a physical peaking resource to meet the extra demand and lower fixed cost instead of holding a swap for that peaking volume.

• The spot price uncertainty is correlated between nearby settlement periods but after about 3 weeks the correlation is negligible because this is about the maximum recovery time for most contingencies that effect supply/demand balance in the short-term. For example, the correlation of average weekly prices in NSW after removing the long-term and seasonal trends is shown in Figure 2-3. The chart shows that for 2 weeks or more separation in time the correlation is negligible. The correlation for adjacent weeks is 33.8% and for two weeks apart it is 10.8%. This means we can evaluate price risk over a period of about a fortnight when considering volume risk. To simplify the modelling we have evaluated risk on a quarterly basis but confined the analysis of correlation to settlement periods within 4 hours of each other. The method for modelling the extent of correlation of spot prices between time periods is described in Appendix B

• The most extreme risk would occur if high volumes were always associated with high prices because retailers are more likely to be:

• under-contracted at times of high demand when they are exposed to inadequate difference payments in their favour; and,

• over-contracted at times of low demand when they are exposed to additional difference payments to generators that are not favourable, especially with negative prices.


Figure 2-3 Correlation of Weekly Average NSW Pool Prices 1997 – 2006

Correlation of Average Weekly NSW Price

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

0 4 8 12 16 20 24 28 32 36 40 44 48 52

Separation in Time (Weeks)

Cor

rela

tion

• A conservative analysis assumes that price and demand are fully monotonic as shown in Figure 2-4. This would over-state the risk and result in over-estimating the optimal contracting level because there would be a strong incentive to fully cover the peak demand with a contract position. In reality, the exposure is not as great because the highest prices do not always occur at the time of highest demand because generators aim to maximise plant availability at such times. Holding a high level of contract could be disadvantageous to a retailer unless the whole retail market also took such a conservative position.

Figure 2-4 Conservative Monotonic Price Versus Demand

Price

Load


For the purposes of assessing an optimal contracting strategy, we have adopted a methodology that:

• takes the spot prices estimated from a PLEXOS analysis for each quarter and assumes they can be modelled as a log normal distribution for each settlement period

• derives a regression relationship between the percentile level of the net system load and the expected price in off-peak periods and peak periods separately assuming the log normal price distribution

• combines the uncertainty due to the mismatch between net system load and expected price with the volatility of the price irrespective of load derived from the PLEXOS as described in Appendix C .

• evaluates an optimal contract position using the average contract price using the method described in Appendix B .

An example of how this method was applied to the 2005/06 historical data is described in Appendix B

2.7 Contract Prices Forward contract prices were taken from historical data acquired by MMA from NGeS and from the d-cypha website on 22 November 2006 and are summarised in the left hand panel of Table 2-3. The Table shows a legend to indicate the source of the data. We have estimated trends in contract prices based upon linear extrapolation and also estimated possible variations away from the trend at a 90% confidence level as discussed below in this section. The annual average contract prices have been split into quarterly prices based upon the ratios of quarterly prices currently quoted by d-cypha.

For the purposes of optimisation of contract positions we have also estimated the 90% firm contract price that would be applicable when purchasing in the year prior to trading for the purposes of fixing the final contract volume. This is based on the proposition that the total contract volume would depend on the contract price applying when purchasing the final tranche. The price of the marginal contract tranche is shown in the right hand panel of Table 2-3. The central panel shows the average contract purchase assuming that the contract position has been acquired from mid 2006.

The future roll-off of ETEF has been an uncertain feature of the wholesale trading environment for the government owned retailers in NSW. It has not been possible for EnergyAustralia to acquire a hedge position for 2008/09 commencing in 2005/06 as appropriate by a prudent retailer. Therefore for the purposes of estimating wholesale costs for a government owned retailer, the mix of historical contract costs must recognise that the portfolio would be acquired from mid 2006. In setting a regulated tariff in common for government owned and private retailers this relatively short notice for the ETEF roll-off imposes additional wholesale costs upon government owned retailers that are subject to ETEF obligations. In this analysis the impact of this constraint has increased the average flat contract price by about $1.93/MWh higher for 2007/08 and $0.30/MWh


Table 2-3 Analysis of Contract Price data - Acquiring Contract Position from August 2006 – updated as at 22 November 2006. Financial Year of Trading ► 2004/05 2005/06 2006/07 2007/08 2008/09 2009/010 2007/08 2008/09 2009/010 2007/08 2008/09 2009/010

Source Legend ► From NGeS datad-cypha

22 Nov 06 Inferred Average Contract PriceInferred Maximum Contract Price (Including

12.6% Options for Final Tranche)Marginal Contract Price (Last Tranche

Including 12.6% Options for Final Tranche)

Contract Product ▼

Calendar Year ▼

Active Period ▼

Flat 2007 36.97 38.08 41.352007/08 37.30 38.10 40.60 40.60 40.90 41.51

2008 37.63 38.12 40.65 41.822008/09 38.60 39.50 40.77 39.92 40.44 41.59

2009 39.08 39.40 39.72 40.042009/10 37.35 37.35 37.35 37.35 39.42 41.03

2007 Q3 37.40 37.40 37.68 38.24Q4 38.00 38.00 38.28 38.85

2008 Q1 54.85 54.85 55.33 56.28Q2 33.400 33.40 33.65 34.15Q3 37.000 37.40 37.88 38.96Q4 38.100 38.51 39.01 40.12

2009 Q1 51.500 52.05 52.79 54.42Q2 31.500 31.84 32.25 33.17Q3 37.350 37.35 38.01 41.03Q4 37.450 37.45 38.12 41.14

2010 Q1 49.750 49.75 50.70 54.85Q2 33.450 33.45 34.04 36.74

Peak 2007 52.76 55.24 61.452007/08 53.19 55.10 61.98 61.98 62.44 63.36

2008 53.62 54.95 62.54 65.962008/09 56.05 56.1 59.53 57.24 57.44 59.07

2009 57.15 55.13 53.10 51.082009/10 54.69 54.69 54.69 54.69 57.71 60.07

2007 Q3 52.00 52.00 52.39 53.16Q4 58.00 58.00 58.43 59.30

2008 Q1 94.15 94.15 94.97 96.61Q2 43.75 43.75 44.08 44.73Q3 53.75 54.85 55.01 56.59Q4 58.50 59.69 59.87 61.59

2009 Q1 87.00 88.77 89.16 91.94Q2 41.50 42.35 42.48 43.70Q3 49.00 49.00 51.71 53.82Q4 43.00 43.00 15.74 47.23

2010 Q1 78.25 78.25 80.03 86.27Q2 48.50 48.50 53.21 53.28


higher in 2008/09. The 2007/08 peak contract is about $5.20/MWh higher under this limitation.

Examination of the history of 2006 and 2007 calendar year flat swap prices presented in Figure 2-5 shows that contract prices have varied by about ±5% over a two year period within a trend range. Peak swap prices have varied within a similar range. It would be reasonable to infer that typical standard deviation of price variation per annum would be about 3.5% on this basis. Therefore using the current contract prices as starting point we could infer that one year ahead contracts could increase by 4.5%, by 6.4% after two years and by 7.8% after three years based on a simple random walk principle. For the purpose of this analysis, we took 2 months, 14 months and 26 months as the lead time to final trading with price scale factors as shown in Table 2-4.

Table 2-4 Risk margin factors for variations in forward contract prices

For 2007/08 For 2008/09 For 2009/10

2 months ahead 1.014

14 Months ahead 1.038

26 months ahead 1.074

The current traded price was multiplied by these factors to represent the traded price when the final contract position is settled. The reasoning was that November 2006 contract prices were used as input and on average it would be expected that 2007/08 financial year prices would be settled by the end of December 2006 on average. The time between November and end December is 2 months. For example, the current swap price for 2009/10 was multiplied by 1.074 to estimate what it could cost when purchased in December 2008 at a 90% confidence level. Based on the NSW spot price forecast in these later years, the contract prices after adjustment for potential price variation did seem to be well below the likely future market prices and it was expected that the NSW contract prices would move up as time advances as they have done in the past as shown in Figure 2-5. The recent rise in the 2007 calendar year contract price is a good example of what can happen leading up to the close of the trading period. This is reasonable because retailers take a risk in purchasing their contract cover forward as the market can move against them. They will be prepared to purchase at higher prices closer to the trading period if necessary to finalise the contract cover.


Figure 2-5 History of Contract Price Trends

CY 2006 Flat swaps

25

27

29

31

33

35

37

39

41

43

45

May-02

Aug-02

Nov-02

Feb-03

May-03

Aug-03

Nov-03

Feb-04

May-04

Aug-04

Nov-04

Feb-05

May-05

Aug-05

Nov-05

VICNSWQLDSA

CY 2007 Flat swaps

25

27

29

31

33

35

37

39

41

43

45

Jun-0

3

Sep-03

Dec-03

Mar-04

Jun-0

4

Sep-04

Dec-04

Mar-05

Jun-0

5

Sep-05

Dec-05

Mar-06

Jun-0

6

Sep-06

Dec-06

VICNSWQLDSA

CY 2008 Flat swaps

25

27

29

31

33

35

37

39

41

43

45

May-04

Aug-04

Nov-04

Feb-05

May-05

Aug-05

Nov-05

Feb-06

May-06

Aug-06

Nov-06

Feb-07

May-07

Aug-07

Nov-07

VICNSWQLDSA


2.7.1 Previous analysis in August 2006

The original analysis was undertaken in August 2006 and the analysis was reviewed using the contract prices that were offered on 22 November 2006 on the d-cypha website. After reducing the premiums for the shorter time period, the contracts that were assessed experienced a 2-3% increase in 2007/08 swap prices, 1-2% increase in 2008/09 prices and a 0-0.6% reduction in prices for 2009/10 against our earlier August assessment. The revised values are shown in Table 2-3.

2.8 Demand Growth Uncertainty and the use of Swap Options EnergyAustralia has examined some of the drivers of peak demand in terms of the penetration of air-conditioning and the economic growth. It has been estimated that 1% increase in the penetration of air-conditioners (currently about 60% in its network service area) would increase peak demand by 20 MW at 1.6kW per customer. Similarly a 1% increase in economic growth would increase peak demand by about 60 MW.

These uncertainties would be reflected in the uncertainty of the estimates of the sales profile by a retailer and would drive the level of option purchase instead of swap and cap positions. As this is a generic assessment, it has been assumed that a stable retail portfolio is served with volume risk related to observed churn rates in the market.

2.8.1 Customer Churn

Mass market customer churn in NSW has grown from 5% of the customer base in 2004/05 to an annualised rate of 8.5% from January to June 2006, equivalent to approximately 300,000 per annum. This is moving closer to the much higher churn rates in Victoria and SA, which have averaged 20%, and may be expected to continue.

MMA’s analysis of reported retailer customer numbers in June 2004 and June 2005 shows that the three incumbents lost up to 20,000 customers or 2.5% of their customer bases i.e. their net losses were approximately half their gross losses. New entrants gained from a few thousand customers (the small new companies such as JackGreen and EnergyOne), to 75,000 customers (large companies new to the NSW electricity market, such as AGL). For these companies these gains represent 30% to 100% growth or more.

With the higher churn rate of 8.5% going forward, incumbents’ gross losses will be higher and for the next three to five years it is reasonable to assume that their net losses will also be higher – at 50% net to gross their losses would be up to 34,000 per year, or 4.2%. After 5 years market shares may stabilise and net losses would reduce, even though churn would continue.

Higher churn represents further gains for existing new entrants but probably diminishing percentage growth rates. Other, brand new, entrants would of course experience very high initial growth.


2.8.2 Treatment of sales uncertainty

Assuming that a new entrant retailer experiences the industry average and recovers half what they lose or loses half what they gain then an allowance of 4.2% per year would be a reasonably conservative range for demand uncertainty. We therefore assume that 4.2% of the contracted volume would be either priced as puts to cover the possible loss of customers and 4.2% calls to represent possible gains of customers. This is treated as an added cost of hedging as discussed above in section 2.8. Since this is traded in the prior year when one third of the contract capacity is purchased, this represents 12.6% of the final contract tranche.

Based upon the d-cypha prices for swap options, the premiums for these products when priced at the strike price level where the put and call options are equally priced is about 1.7% for 2007 swap options, 6.3% for the 2007 Q1 swap option, 13.8% for the 2008 Q1 swap option and 17.8% for the 2009 Q1 swap option. This is equivalent to 6.4% per year lead time when purchased in the year ahead. No information was available on swap options for the other quarters, so the annual premium was applied as a conservative estimate. The option premiums for the Q1 swap and the calendar year swap suggest that nearly all the risk is assessed to be Q1. For the purpose of this analysis, a premium to the swap and cap contract price was applied for the purchase of swap options as shown in Table 2-5.

Table 2-5 Premiums to allow for 12.6% purchase of swap options for final swap tranche

Option for: Q1 swap and cap

Q2 – Q4 swap or cap

Estimated option fee 6.4% 1.7%

Price premium applied to final contract volume (representing 12.6% of contract volume) 0.81% 0.21%

2.9 Composite of Contract Prices In Table 2-3 we have presented an estimate of an upper bound of average contract prices for the three financial years by either applying the above risk premiums and option premiums according to the time frame from a starting price (for example 5.4% for the 2009/10 price in 2006/07) or applying them to the trend line where that is evident as for 2009 calendar year prices. This risk of increasing contract prices and the impact of inflation counteracts the current declining trend and gives a fairly stable net result with an average contract price of $40.25/MWh for flat contracts and $59.20/MWh for peak contracts over the three years. This seems reasonable given the development of Tallawarra will slightly increase competition in NSW which should balance the escalation pressure on prices. The application of the inferred maximum average contract prices shown in Table 2-3 is recommended by MMA as the basis for product pricing and the marginal contract price as the basis for portfolio sizing.


Table 2-6 Forecast $300 Cap Prices Financial Year of Trading ► 2004/05 2005/06 2006/07 2007/08 2008/09 2009/010 2007/08 2008/09 2009/010 2007/08 2008/09 2009/010

Source Legend ►NGeS 21 Aug 06

d-cypha 22 Nov 06 Inferred Average Contract Price

Inferred Maximum Contract Price (Including 0.0% Options for Final Tranche)

Marginal Contract Price (Last Tranche Including 0.0% Options for Final Tranche)

Cap $300 2007/08 10.76 10.76 10.86 11.042008 11.55

2008/09 11 11.00 11.28 11.622009/10 11 11.00 11.62 12.13

2007 Q3 7.25 7.25 7.31 7.44Q4 10 10.00 10.09 10.26

2008 Q1 21.8 21.80 21.99 22.37Q2 4 4.00 4.03 4.10Q3 8.4 7.41 7.60 7.83Q4 12 10.22 10.48 10.80

2009 Q1 22.28 22.84 23.55Q2 4.09 4.19 4.32Q3 7.41 7.83 8.17Q4 10.22 10.80 11.27

2010 Q1 22.28 23.54 24.56Q2 4.09 4.32 4.51


2.10 Cap Prices There was little traded history on $300 cap prices for NSW and based on this scant information, MMA developed a price forecast as shown in Table 2-6. The data was sourced from NGeS and some obtained from d-cypha. The allocation of caps to quarterly prices was based upon the current ratios indicated by the d-cypha prices for 2007/08. We would expect that incumbent generators will continue to withdraw capacity to support market prices in response to Tallawarra and that price volatility would not be greatly affected. The future uncertainty of the cap prices and the application of cap options to manage volume uncertainty have been estimated using the same parameters as for peak swaps.

2.11 NGAC Prices The NSW Greenhouse Gas Abatement Scheme requires retailers to purchase abatement certificates so that their total Greenhouse gas emissions achieve a target related to tonnes of CO2 per head of population which varies by year. The market for NGACs is currently under-supplied and prices have risen to the cap price. Just recently, NGAC prices have fallen due to a large number of NGACs that have been produced by retailers and intermediaries giving away compact fluorescent light globes to earn NGACs. The NGAC allowance per light globe is likely to be reviewed by IPART because there is some survey evidence that not all the globes issued have been installed and displaced incandescent globes.

This recent development illustrates the difficulty faced in estimating future NGAC prices. The price is likely to be volatile as new sources of abatement are brought to market. The operation of Tallawarra will provide a new source of NGACs which would be expected to lower the price. Queensland generators are also eligible to produce NGACs rather than Gas Electricity Certificates (GECs) for the Queensland Scheme and would do so if the NGAC dollar per MWh of generation were higher than the GEC price per MWh of generation. This decision would be likely made by combined cycle generation and cogeneration in Queensland because the NGAC value per MWh is higher for the less GHG intensive generation plant.

A prudent retailer facing this scenario would need to contract in the longer term for NGACs and this does require new abatement capacity to come on line. On that basis we argue that the appropriate NGAC price would be based upon the NGAC subsidy required from a new combined cycle gas fired plant operating at an efficient load factor of 42% capacity factor to be competitive with a new base load super-critical coal fired plant operating at 90% capacity factor. Using the cost parameters provided in Table 2-7, we estimate the expected NGAC price at $12.79/NGAC as shown in the same Table. Allowing for uncertainty in the cost parameters, we determine a long-term firm NGAC price in 2006 dollars of $14.15/NGAC. This price is escalated at CPI – 1% to reflect long-term trends in electricity production costs.


This price is below that which has been traded in recent times and is above the original estimate of long-term NGAC prices of $8 estimated by Frontier Economics in 2002. We expect NGAC prices to moderate as new supply responds to the extension of the Scheme to 2020.

A prudent retailer would have been purchasing its requirements in prior years and would not purchase its NGAC requirements on the run. Therefore, it is appropriate to examine traded prices for NGACs from prior years and formulate a purchasing plan that either:

• locks in forward contracts; or

• purchases NGACs on the spot market and holds them for acquitting in future periods.

Using this method we formulate an annual average cost as illustrated in Table 2-9. We take the minimum of the forward price and the holding price allowing 3% for CPI and 5% risk margin for holding the NGACs as a forward purchase. The total mark-up for holding NGACs is therefore 8.15% per annum. In most cases forward contracting is preferred over purchase and hold.

This gives us estimated NGAC prices of $14.87 in 2007/08, $15.65 in 2008/09 and $15.01 in 2009/10 in nominal dollars as seen at the bottom of Table 2-9. In the NGAC spot market, these prices could vary much more considerably. However, a prudent retailer should have been able to lock in most of its NGAC obligation at these values or less by establishing a forward contracting strategy over the last three years.

2.12 Mandatory Renewable Energy Target The Commonwealth Government’s policy, MRET, aims to achieve 2% additional renewable energy by 2010. It has been implemented as a 9,500 GWh target with a maximum penalty for non-performance of $40/MWh. This penalty is not indexed to CPI. The penalty is also not tax deductible, meaning that under current company tax rates a liable party would be indifferent between paying the penalty or purchasing certificates at a price of $57/MWh. This penalty would effectively provide a cap on the premium available for renewable energy. Whilst a ramp-up target schedule has been developed for each calendar year by the Government as shown in Table 2-10, a credits banking regime will stimulate earlier development of such projects.

2.12.1 Renewable Power Percentage of Wholesale Purchases

The retailer percentages for each of the next four calendar years have been estimated by comparing total wholesale sales with the MRET targets applicable to the corresponding years as shown in Table 2-8. The ratios have been estimated assuming 3%pa energy demand growth over the next four years.


Table 2-7 New Entry Costs for the Purpose of Estimating LRMC of NGACs

Parameter Units New coal fired technology

Low emission technology Uncertainty Measure Basis Comments

Overnight Capital Cost $/kW $/kW $1,500 $1,000 10% Standard Deviation of Cost

Economic Life Years Years 30 25 5 Range of Years Uniform Distrbution

Interest During Construction % real 9.29% 7.67% 10% Standard Deviation of ratio

Based upon 5% real interest rate

Marginal Transmission Loss Factor 0.9756 0.9887 0.02 Uniform Range CCGT at Munmorah,

Coal at Mt Piper

Fixed Operating Cost $/kW/year $/kW/year $35 $20 20% Standard Deviation of Cost

Variable Operating Cost $/MWh $/MWh $2 $3 20% Standard Deviation of Cost

Variable Fuel Cost $/GJ $/GJ HHV $1.40 $3.90 10% Standard Deviation of Cost

Based on MMA-Gas Model

Heat Rate GJ/MWh GJ/MWh 9.2 7.3 2% Standard Deviation of Value

Fuel CO2 Rate kg/GJ kg/GJ 90 56 2% Standard Deviation of Value

Efficient Dispatch Capacity Factor without abatement % 90% 42.5% 2% Range of Values

for Coal only CCGT by Calculation

Average off-peak revenue for Coal Plant for 20% capacity factor $/MWh

$/MWh $20 Not Applicable 10% Standard Deviation of Value

Weighted Average Cost of Capital (Real Pre-Tax) Real Pre-Tax 7.80% 7.80% 0.5% Absolute Range Uniform Distrbution

Base Levelised Cost to Pool for 42.5% Capacity factor $/MWh $/MWh $58.92 $64.23 By calculation

Standard Deviation of Cost $/MWh $/MWh $5.97 $6.33 By calculation

NGAC Subsidy to equalise costs $/MWh $/MWh $5.31 To equalise costs

Standard Deviation of NGAC Subsidy $/MWh $/MWh $0.63

Emission Rate t/MWh t/MWh 0.828 0.4088Base NGAC Price $/t CO2 $/t CO2 $12.66Mean of 10,000 samples $/t CO2 $12.79Standard Deviation of 10,000 samples

$/t CO2 $1.62

80% Confidence Level $/t CO2 $14.15

Table 2-8 Renewable Energy Certificate Factors

Calendar Year 2006 2007 2008 2009 2010

Total Sales (GWh)

207,373 213,594 220,002 226,602 233,400

MRET (GWh) 4,500 5,600 6,800 8,100 9,500

Ratio (%) 2.17% 2.62% 3.09% 3.57% 4.07%


Table 2-9 Calculation of NGAC Price for Future Years (Nominal Price)

2004/05 2005/06 2006/07 2007/08 2008/09 2009/10

New NGACs $14.15 $14.43 $14.72 $15.01

Average TradedNGAC Price

2004/05 11.43 12.23 12.93 13.57

2005/06 12.87 14.63 15.29 15.73

2006 15.76 16.02 16.49

Holding Cost at 5%Real Cost for

From 2004/05 Holding 12.34 13.21 13.97Forward Price 12.23 12.93 13.57Minimum 12.23 12.93 13.57

From 2005/06 Holding 13.90 15.02 16.22Forward Price 14.63 15.29 15.73Minimum 13.90 15.02 15.73

From 2006/07 Holding 17.02 17.30Forward Price 16.02 16.49Minimum 16.02 16.49

From 2007/08 Forward Price 14.72 15.01From 2008/09 Forward Price 15.01Average Cost ofNGACs 12.76 14.87 15.65 15.01

Table 2-10 Renewable energy targets Calendar Year Target (GWh )

2002 1100

2003 1800

2004 2600

2005 3400

2006 4500

2007 5600

2008 6800

2009 8100

2010 and later years 9500


2.12.2 Renewable Energy Certificate Prices

The price of RECs in the spot market has fallen markedly since 2003 due to a looming over-supply of RECs caused by a substantial contribution from solar hot water heaters and sufficient commitments of new renewable energy generation to meet the 2010 MRET of 9,500 GWh. There are sufficient resources to meet the target through to 2020 assuming that committed plants continue to operate and existing hydro power stations operate above their baselines at recent levels.

The trading in spot RECs is quite thin as new renewable generators need to contract their supply long-term to make their projects bankable. This means that the current spot prices are not a suitable indicator of the costs that would be incurred by a prudent new entrant retailer in sourcing substantial volumes of RECs. It is more appropriate to estimate average cost of RECs rather than the marginal cost of RECs as indicated by current spot trading because the great majority of RECs have been sold under long-term contracts.

For this purpose we compare the typical cost of new renewable energy projects with corresponding energy prices in the regions of Australia with higher energy costs and/or lower cost renewable energy resources. This can be done effectively using MMA’s renewable energy market model which evaluates the optimal sourcing of RECs throughout Australia. The method and data applied are described in detail in Appendix A . This gives a long-run average REC price of between $38 and $44/REC in nominal dollars as shown in Table 2-11. The price for 2007/08 is influenced by the lower spot prices during 2005 and 2006. The price for 2009/10 is based on the longer term average REC cost in nominal dollars.

The information in Table 2-11 makes it apparent that the cost of RECs could range from about $27 to some $40 depending on when the retailer takes up its REC position. The values estimated in Table 2-11 represent a middle range outcome based on the three year trading view. Recent spot market trades of RECs which have fallen to $19 in September 2006 and $17 in November 2006 are not particularly relevant for this purpose because the volume of RECs traded at this price is minimal and retailers will have needed to contract long-term to acquire the necessary RECs so as to support financial commitment to the renewable energy projects.


Table 2-11 Analysis of REC Prices (Nominal Dollars)

2004 2005 2006 2007 2008 2009 2010Average Contract

REC Price ►$38.19 $41.98 $43.24 $44.53

Average Traded RECPrice ▼2004 (Dec) 39.4 38.75 40.83 43.47 46.13 47.672005 (Dec) 38.73 26.9 27.75 29.25 31.25 32.252006 (June) 26.9 20.5 22.75 24.5 25.75Holding Cost at 5%Real Cost forFrom 2004 Holding 42.55 41.85 44.10 46.95 46.95

Forward Price 38.75 40.83 43.47 46.13 47.67Minimum 38.75 40.83 43.47 46.13 46.95

From 2005 Holding 41.83 29.05 29.97 29.97 31.59Forward Price 26.90 27.75 29.25 31.25 32.25Minimum 26.90 27.75 29.25 29.97 31.59

From 2006 Holding 29.05 22.14 31.38 23.91Forward Price 20.50 22.75 24.50 25.75Minimum 20.50 22.14 24.50 23.91

From 2007 Forward Price 41.98 43.24 44.53From 2008 Forward Price 43.24 44.53Average AnnualCost of RECs 30.57 31.12 36.99 37.66

2007/08 2008/09 2009/10Average Fin YearCost of RECs 30.85 34.06 37.32

2.13 NEMMCO Fees

The historical trend in NEMMCO fees is shown in Figure 2-6. We assume that the historical trend will continue over the future period for each of the fee components individually. Unit fees are decreasing as the cost of establishing the NEM and the systems for Full Retail Contestability are amortised over 10 years and as the market grows in size. Table 2-12 shows the estimate of NEMMCO fees for the period by extrapolation. Of particular note is that the establishment fees for establishment of retail contestability and the NEM systems are decreasing significantly due to the 10 year amortisation. Since these costs are a minor part of the total cost structure, we do not consider that any more detailed analysis is warranted.


Figure 2-6 Trend in NEM Fees (Nominal Dollars)

Trend in NEM Fees

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

2005 2006 2007 2008 2009 2010

Financial Year Ending June

$/M

Wh

Full Retail CompetitionOperatingFull Retail CompetitionEstablishmentTotal NEMMCO

Table 2-12 NEM Fees

2006/07 2007/08 2008/09 2009/10

NEMMCO 0.2551 0.25466 0.256 0.25733

Retail Contestability Establishment

0.043 0.03307 0.02135 0.00962

Retail Contestability Operations

0.05761 0.05801 0.05696 0.0559

Total 0.35571 0.34575 0.3343 0.32286

2.14 Ancillary Services Costs

Ancillary services costs are only a small part of the overall costs of energy in the spot market. They average around 0.6% of energy costs. When the current ancillary service regime was established there were some very high costs incurred due to some transmission outages and some high bids from participants not fully appreciating the new


Figure 2-7 History of Customer Ancillary Service Cost $/MWh

NEM Customer Ancillary Services Cost

$0.00

$0.20

$0.40

$0.60

$0.80

$1.00

$1.20

$1.40

7/03

/200

4

7/04

/200

4

7/05

/200

4

7/06

/200

4

7/07

/200

4

7/08

/200

4

7/09

/200

4

7/10

/200

4

7/11

/200

4

7/12

/200

4

7/01

/200

5

7/02

/200

57/

03/2

005

7/04

/200

5

7/05

/200

5

7/06

/200

5

7/07

/200

5

7/08

/200

5

7/09

/200

5

7/10

/200

5

7/11

/200

5

7/12

/200

5

7/01

/200

6

7/02

/200

67/

03/2

006

7/04

/200

6

7/05

/200

6

7/06

/200

6

7/07

/200

6

$ pe

r MW

h d

eman

d

REVISE FINAL PRELIM

scheme. Figure 2-7 shows the trend in total ancillary service costs that are applicable to energy purchases since March 2004 as provided on the NEMMCO website.

It is apparent that unit costs have been decreasing as interconnections have been upgraded and new generating plant has entered service. Previous estimates for long term ancillary costs were $0.56/MWh in the CRA report of December 2003. Extrapolating the historical trend apart from the initial spike on a linear basis is not helpful because it yields zero cost by June 2009. Extrapolating from July 2005 gives $0.13/MWh by mid 2009. There is no clear basis to predict any continuing decline in cost, so a constant nominal cost of $0.30/MWh would seem reasonable.


3 WHOLESALE COST ANALYSIS

3.1 Long Run Marginal Cost The analysis of LRMC provided estimates of mean, standard deviation and 80% firm value for LRMC for the three years based upon the shape assessed for the future net system load profile. The price has been assessed for the 50% POE peak demand to represent a median value as shown in Figure 3-1 for 3% inflation. The LRMC for 2005/06 was based on the 2005/06 net system load profile.

The chart shows the nominal price with and without allowing for the revenue obtained from emission abatement through the NSW GGAS. It is recommended that this revenue should be included to avoid double counting the costs of energy and NGACs. The LRMC with abatement credit declines from 2005/06 to 2007/08 due to an assumption that NGAC certificate prices will increase in the next two years. From 2007/08 onwards, the rising price trend is due to the assessed declining load factor of the net system load due to increased heating and cooling load.

Figure 3-1 Long-Run Marginal Cost for 50% POE Peak Demand

LRMC (50% POE)

$55.0

$57.0

$59.0

$61.0

$63.0

$65.0

$67.0

2006 2007 2008 2009 2010


$/M

Wh

Nom

inal

No Abatement CreditWith Abatement Credit

Based upon the cost of base load power at 100% utilisation, the ratio of volume weighted price to base load price was also estimated as shown in Figure 3-2. The price ratio according to the LRMC price model is quite stable at about 1.40. The price ratio obtained from the PLEXOS results is much lower at between 1.10 and 1.15. The lower assessed


value reflects the difficulties of representing infrequent and extreme pool price spikes in PLEXOS. Since such spikes are an essential feature of an energy only market without a fully developed demand side response, we would expect that historical volume ratios

Figure 3-2 Ratio of Volume to Time Weighted Price

Volume Cost ratio

0.800

0.900

1.000

1.100

1.200

1.300

1.400

1.500

2006 2007 2008 2009 2010


LRMC CostRatio

PLEXOS SpotCost Ratio

would be more applicable than those obtained from the PLEXOS model. Off-peak prices are expected to rise after service of Basslink and indeed have so substantially and therefore some reduction in the ratio of volume weighted price to time weighted price is expected but not quite to this extent.

3.2 Contract Coverage Using the methodology described in Chapter 2 and Appendix B and Appendix D , the flat and peak contract volumes were estimated for the net system load.

In some cases there was a mismatch of up to 25% between the expected spot prices and the assessed contract price so that the calculations suggested that a retailer would contract much less if the contract price were high or much more if the contract price were low. This would not occur in practice because the small risk of extreme events not captured in the analysis requires a retailer to substantially match the energy traded. In this analysis the trading volume was estimated by bringing the contract price into alignment with the spot market modelling so that an optimal contract volume could be obtained (rather than zero or infinity) as shown in Figure B- 7 of Appendix B .

An example of the load profile that was forecast and the peak and off-peak swap volume is shown in Figure 3-3. It may be observed that there remains substantial volume exposed to the spot market on the weekends and that in mid January and towards the end of


March even the peak period is over-contracted. This suggests that there would be scope for sculpting the contract volume during the quarter, but the savings from such adjustments are likely to be offset by the spot exposure to a substantial degree. The


Figure 3-3 Load and Swap Volume for January to March 2009

Jan 2009

0.000

0.200

0.400

0.600

0.800

1.000

1.200

1.400

1.600

1.800

2.000

1-Jan-09 4-Jan-09 7-Jan-09 10-Jan-09 13-Jan-09 16-Jan-09 19-Jan-09 22-Jan-09 25-Jan-09 28-Jan-09 31-Jan-09Settlement Period Ending

MW

h R

egul

ated

Loa

d

LoadContract for Minimum Error

Feb 2009

0.000

0.200

0.400

0.600

0.800

1.000

1.200

1.400

1.600

1-Feb-09 4-Feb-09 7-Feb-09 10-Feb-09 13-Feb-09 16-Feb-09 19-Feb-09 22-Feb-09 25-Feb-09 28-Feb-09Settlement Period Ending

MW

h R

egul

ated

Loa

d

Load

Contract for Minimum Error

Mar 2009

0.000

0.200

0.400

0.600

0.800

1.000

1.200

1.400

1-Mar-09 4-Mar-09 7-Mar-09 10-Mar-09 13-Mar-09 16-Mar-09 19-Mar-09 22-Mar-09 25-Mar-09 28-Mar-09 31-Mar-09Settlement Period Ending

MW

h R

egul

ated

Loa

d

Load

Contract for Minimum Error


excess volumes are covered by the cap contract above $300/MWh, so the overall exposure is acceptable as discussed in section 3.3. It is apparent that the optimal contract level in the off-peak period is well above the minimum load even on weekdays. The chart shows that the over-night load on weekdays is normally more than fully covered by the flat contract component.

The ratio of the flat contract volume to the average off-peak load is shown in Figure 3-4 from 2007/08 to 2009/10.

Figure 3-4 Swap Contract Coverage

Ratio of Swap Contract Volume to Average Load

94%

96%

98%

100%

102%

104%

106%

108%

110%

112%

Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2

2007 2008 2009 2010

Quarter

Off-Peak PeriodPeak Period

There is a consistent pattern of variation from quarter to quarter. The overall result is that the optimal contract coverage in the off-peak period between 3% and 10% above the average load level. The required swap cover in peak periods also varies from the average peak period load by up to 10%. A higher level of contracting is viable in Q2 when load levels and price volatility are lower than in the other quarters of the years. These ratios appear to be sensitive to the actual weather pattern that is modelled. There is more consistency in the profile for the off-peak period, perhaps because the load variation in off-peak periods is less affected by weather conditions, except at weekends which are included in the off-peak period.

3.3 Wholesale cost The wholesale cost was obtained by adding up the cost of the contracts and the spot market exposure at a 90% probability that the value would exceed the spot market cost


exposure. The analysis of costs for the load profiles that were developed is shown in Table 3-1.

Table 3-1 Analysis of Wholesale Cost Quarter 2007 2008 2009 2010

Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2LRMC Price $/MWh $56.53 $56.53 $56.53 $56.53 $60.09 $60.09 $60.09 $60.09 $62.03 $62.03 $62.03 $62.03

Wholesale Price $/MWh $44.97 $52.98 $87.16 $40.73 $46.38 $54.08 $90.71 $39.13 $46.57 $50.48 $90.75 $41.77

With Emission Abatement and NEM Services $/MWh $50.05 $58.24 $92.75 $46.37 $52.04 $59.79 $96.83 $45.21 $52.74 $56.60 $97.04 $48.01

Peak Hours Hours 960 945 945 945 975 945 930 930 975 945 930 930Off-Peak Hours Hours 1248 1263 1215 1239 1233 1263 1230 1254 1233 1263 1230 1254

Peak Spot Price $/MWh 38.32 34.35 36.67 41.71 37.63 36.11 33.78 46.07 38.17 37.54 35.00 47.14Off-Peak Spot Price $/MWh 28.85 30.32 30.68 28.59 28.81 25.64 31.07 29.49 29.62 26.04 29.22 30.02Average Spot Price $/MWh 33.42 32.26 33.65 34.90 33.13 30.72 32.39 37.51 33.82 31.65 32.04 38.34

NSL+CLPeak Period MWh 1873 1519 1614 1815 1982 1591 1728 1851 2074 1711 1865 1954Off-Peak Period MWh 2005 1639 1642 1956 2068 1687 1827 1976 2149 1794 1956 2065Total MWh 3879 3158 3256 3771 4050 3278 3555 3827 4223 3505 3821 4020

Peak 10% POE MW 3.105 3.329 3.841 3.321 3.431 3.482 4.268 3.476 3.574 3.826 4.695 3.60850% POE MW 3.053 3.316 2.549 3.321 3.368 3.474 3.700 3.476 3.557 3.817 4.073 3.60890% POE MW 3.053 2.862 2.549 3.321 3.431 3.012 3.198 3.469 3.557 3.311 3.523 3.600

Contract VolumesFlat MW 1.687 1.363 1.475 1.694 1.774 1.394 1.627 1.694 1.814 1.468 1.738 1.755Peak MW 0.315 0.298 0.259 0.435 0.315 0.292 0.235 0.497 0.406 0.375 0.275 0.577Cap MW 1.104 1.944 1.779 1.192 1.342 1.797 2.406 1.285 1.354 1.983 2.683 1.275

Contract Volume / Average LoadOff-peak 105% 105% 109% 107% 106% 104% 110% 107% 104% 103% 109% 107%Peak 103% 103% 102% 111% 103% 100% 100% 110% 104% 102% 100% 111%

Average Contract PricesFlat $/MWh 37.68 38.28 55.33 33.65 37.88 39.01 52.79 32.25 38.01 38.12 50.70 34.04Peak $/MWh 52.39 58.43 94.97 44.08 55.01 59.87 89.16 42.48 51.71 15.74 80.03 53.21Cap $/MWh 7.31 10.09 21.99 4.03 7.60 10.48 22.84 4.19 7.83 10.80 23.54 4.32

Contract CostFlat $000 $140.3 $115.2 $176.2 $124.5 $148.4 $120.0 $185.6 $119.3 $152.2 $123.6 $190.3 $130.5Peak $000 $15.83 $16.47 $23.24 $18.13 $16.92 $16.52 $19.47 $19.65 $20.49 $5.58 $20.45 $28.56Cap $000 $17.82 $43.30 $84.52 $10.50 $22.50 $41.56 $118.69 $11.76 $23.40 $47.27 $136.44 $12.03

Spot ExpectedOff-Peak $000 -0.224 -0.418 -0.613 -0.393 -0.781 -0.321 -0.204 -0.777 -0.231 0.023 -1.331 -1.047Peak $000 0.009 -0.033 -0.272 -0.147 0.133 -0.110 0.015 -0.670 -0.099 -0.275 0.026 -1.769Cap $000 -0.001 -$6.90 -$0.00 -$0.06 0.000 -$0.08 -$0.28 -$0.32 -$0.00 -$0.00 -$0.01 -$0.95

Risk Margin on Spot ExposureOff-Peak $000 $0.27 $0.26 $0.41 $0.46 $0.33 $0.26 $0.89 $0.49 $0.37 $0.34 $0.70 $0.56Peak $000 $0.41 $0.32 $0.25 $0.82 $0.33 $0.32 $0.24 $2.12 $0.59 $0.45 $0.31 $4.03Cap $000 -$0.01 -$0.92 -$0.01 -$0.26 -$0.00 -$0.92 -$1.94 -$1.79 -$0.05 -$0.02 -$0.12 -$4.01

Total Firm Market CostFlat $000 $140.4 $115.1 $176.0 $124.6 $148.0 $120.0 $186.3 $119.0 $152.4 $123.9 $189.7 $130.0Peak $000 $16.24 $16.75 $23.23 $18.80 $17.38 $16.73 $19.73 $21.10 $20.98 $5.76 $20.79 $30.82Cap $000 $17.81 $35.47 $84.51 $10.18 $22.50 $40.56 $116.47 $9.65 $23.35 $47.26 $136.32 $7.07

Peak Period $000 $95.06 $101.55 $184.84 $82.86 $105.40 $108.67 $216.10 $81.54 $111.55 $105.90 $239.04 $93.45Off-Peak Period $000 $79.36 $65.76 $98.94 $70.70 $82.42 $68.61 $106.36 $68.20 $85.14 $71.05 $107.74 $74.44

Total Energy Market Cost $000 $174.42 $167.31 $283.77 $153.56 $187.82 $177.27 $322.46 $149.75 $196.69 $176.95 $346.78 $167.89

ETEF Share 100% 100% 100% 100% 93% 80% 73% 60% 53% 40% 33.3% 20%Note: Reflects costs at NSW RRN, excludes distribution and transmission losses.


The Table shows from the top:

• The LRMC is based on the new entry costs, their uncertainty and the shape of the net system load

• The wholesale price is based on optimal spot and contract market trading excluding emission abatement costs and ancillary costs and system losses;

• The wholesale cost with emission abatement and ancillary service costs;

• The peak and off-peak hours allocated to each quarter;

• The peak and off-peak price obtained from the PLEXOS solution;

• The net system load in peak and off-peak periods based on 1 MW initial average consumption on weekends. This load level was arbitrarily applied and has no bearing on the results which are determined by the load shape over time, not its magnitude;

• The peak demand in the load shape for the variations in weather corresponding to 90%, 50% and 10% probability of exceedance (POE). In some quarters the weather did not affect the peak because the extremes occurred on a weekend. There was no adjustment of the weather data to ensure that extreme weather occurred on weekdays. [please update to include first year];

• The contract load in each quarter measured in MW referred to the NSW Regional Reference Node;

• The average contract price including the effect of accumulation of the contract position over three years, the premium for 12.6% of the contract volume in the last period to allow purchase of options to provide for load volume uncertainty and a provision to recognise the risks that future purchases of forward contracts could be at higher prices based upon contract price volatility;

• The cost of the contract as the product of volume and strike price for swaps and option fee times volume for caps;

• The expected residual exposure to the spot market due to the mismatch between hedge position and load purchase from the pool. This is a price at the reference node in accordance with the spot price forecasts;

• The additional risk to the spot market purchases which represents a 90% probability that the expected cost plus the risk margin will exceed the actual cost based on price and load variations;

• The total firm market cost of the contracts and spot market purchases or sales, allowing for the risk margin. This cost is divided by the energy volume to determine an average price;

• The share of the net system load that is to be supplied from ETEF

The initial analysis for 2007/08 transferred the summer peak from Q1 2008 to late in Q4 2007 because of the way that the temperatures were applied to the days of the week in an


attempt to ensure that the peak day was a working day. However, this produced a lower wholesale cost of about $52.50/MWh for 2007/08. To correct for this anomaly, the 10% POE peak was increased in Q1 2008 and the corresponding 10% POE peak was reduced in Q4 2007 to indicate a more stable pattern from year to year and to ensure that sufficient cap volume was included in the analysis.

The quarterly analysis of the wholesale costs is illustrated in Figure 3-5.

Figure 3-5 Quarterly Prices

Quarterly Average Energy Cost

$0

$10

$20

$30

$40

$50

$60

$70

$80

$90

$100

Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2

2007 2008 2009 2010

Quarter

$/M

Wh

LRMCWholesale Cost

The chart shows:

• The long run marginal cost (LRMC) analysis for the net system load taking the assessment on an annual basis.

• The wholesale cost derived from spot and contract trading including risk margins for variability of the spot and contract market prices for future trading periods.

• The trend of the composite cost is shown to be declining relative to the LRMC because the wholesale cost is expected to remain below the LRMC in the period to June 2010.

The range of wholesale price applicable to the projected evolution of the net system load plus controlled load segment varied from $55.40/MWh in 2007/08 for the wholesale market price to $62.03/MWh in 2009/10 for the LRMC as shown in Table 3-2.


Table 3-2 Annual Wholesale Cost (Energy Only)

Financial Year 2008 2009 2010Ending June Units



ETEF Share 100% 77% 37%

Note: Reflects costs at NSW RRN, excludes distribution and transmission losses.

3.4 Abatement and other costs The addition of the abatement costs for RECs and NGACs increase the wholesale cost by about $5 to $6/MWh over the review period. The analysis of cost components for the volumes shown in Table 3-1 is shown in Table 3-3. The yellow shaded cells show direct relationship to annual or quarterly estimated prices.

The annual summary is shown in Table 3-4. The tables show the cost of the abatement products as well as the share of the retail cost at the wholesale level after the relevant percentage ratios are applied.


Table 3-3 Quarterly analysis of additional abatement cost and NEM services

Energy Abatement and NEM Services

Quarter 2007 2008 2009 2010Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2

Total Energy Cost $000 $174.4 $167.3 $283.8 $153.6 $187.8 $177.3 $322.5 $149.7 $196.7 $177.0 $346.8 $167.9

REC Price $/MWh $30.57 $30.57 $31.12 $31.12 $31.12 $31.12 $36.99 $36.99 $36.99 $36.99 $37.66 $37.66Renewable Power % 2.62% 2.62% 3.09% 3.09% 3.09% 3.09% 3.57% 3.57% 3.57% 3.57% 4.07% 4.07%REC Cost $000 $3.1 $2.5 $3.1 $3.6 $3.9 $3.2 $4.7 $5.1 $5.6 $4.6 $5.9 $6.2

NGAC Price $/NGAC $13.55 $14.21 $14.87 $15.06 $15.26 $15.45 $15.65 $15.49 $15.33 $15.17 $15.01 $14.85NEC Ratio NGAC/MWh 0.268 0.27 0.27 0.27 0.266 0.27 0.27 0.27 0.275 0.28 0.28 0.28NGAC Cost $000 $14.09 $12.03 $12.98 $15.22 $16.45 $13.49 $14.81 $15.78 $17.82 $14.64 $15.79 $16.44

NEM Fee Levy $/MWh $0.35 $0.35 $0.35 $0.35 $0.33 $0.33 $0.33 $0.33 $0.32 $0.32 $0.32 $0.32NEM Fees $000 $1.34 $1.09 $1.13 $1.30 $1.35 $1.10 $1.19 $1.28 $1.36 $1.13 $1.23 $1.30

Ancillary Services Charge $/MWh $0.30 $0.30 $0.30 $0.30 $0.30 $0.30 $0.30 $0.30 $0.30 $0.30 $0.30 $0.30Ancillary Services $000 $1.16 $0.95 $0.98 $1.13 $1.21 $0.98 $1.07 $1.15 $1.27 $1.05 $1.15 $1.21

Total Cost $000 $194.1 $183.9 $302.0 $174.8 $210.7 $196.0 $344.2 $173.0 $222.7 $198.4 $370.8 $193.0

Total Supply Cost $/MWh $50.05 $58.24 $92.75 $46.37 $52.04 $59.79 $96.83 $45.21 $52.74 $56.60 $97.04 $48.01

Additional Cost $/MWh $5.08 $5.26 $5.59 $5.65 $5.66 $5.71 $6.12 $6.08 $6.17 $6.12 $6.29 $6.25


Table 3-4 Annual summary of additional abatement cost and NEM services

Financial Year 2008 2009 2010Ending JuneTotal Energy Cost $000 $779.07 $837.30 $888.32Total Energy Price $/MWh $55.40 $56.92 $57.06

REC Price $/MWh $30.87 $34.28 $37.35RPP 2.86% 3.33% 3.82%REC Charge $/MWh $0.88 $1.14 $1.43REC Cost $000 $12.40 $16.81 $22.24

NGAC Price $/NGAC $14.41 $15.46 $15.09NEC Ratio NGAC/MWh 0.27 0.27 0.28NGAC Charge $/MWh $3.86 $4.12 $4.16NGAC Cost $/MWh $54.32 $60.53 $64.69

NEM Fee Levy $/MWh $0.35 $0.33 $0.32NEM Fees $000 $4.86 $4.92 $5.03

Ancillary Services Charge $/MWh $0.30 $0.30 $0.30Ancillary Services $000 $4.22 $4.41 $4.67

Total Cost $000 $854.86 $923.98 $984.95

Total Supply Cost $/MWh $60.79 $62.81 $63.26

Additional Cost $/MWh $5.39 $5.89 $6.21

3.5 Plexos Modelling Results The PLEXOS model results show that in the next four years NSW spot prices will be around $38/MWh to $40/MWh in nominal terms. Figure 3-6 shows the historical and forecast flat, peak and off-peak spot prices in NSW. PLEXOS forecasts of peak prices are generally lower than peak prices observed in the last two years and revert to the trend of 2003 and 2004 financial years. PLEXOS forecasts of off-peak prices are generally higher than those observed historically due to the impact of Basslink and higher loading of coal fired plant due to demand growth. Figure 3-7 shows the historical and forecast price duration curves from NSW. It can be seen that the PLEXOS model does not exhibit quite the same level of price volatility as observed in the past. This is because of the difficulty in simulating the occasional price spikes that occur due to gaming prices down from the $10,000/MWh value of VoLL. Some increase in off-peak prices is expected to occur due to the impact of Basslink and the consequence that the brown coal plants in Victoria are now infrequently marginal in off-peak periods. The reduced off-peak imports from Victoria allow higher off-peak prices in NSW.

From 2006/07 to 2009/10 the PLEXOS model includes the effects of the full operation of Basslink and new entrants such as Laverton North, Kogan Creek and Tallawarra. The additional generation capacity and interconnector flow from Tasmania will have a dampening effect on peak prices. The operation of Basslink will also cause an increase in


Figure 3-6 Historical and Forecast NSW Pool Prices

NSW Pool Price

$0

$10

$20

$30

$40

$50

$60

$70

2003 2004 2005 2006 2007 2008 2009 2010

Financial Year

Pool

Pri

ce Flat

Peak

Off Peak

Actual Forecast

Figure 3-7 Historical and Forecast NSW Price Duration Curves

NSW Price Duration Curve

$10

$100

$1,000

0% 20% 40% 60% 80% 100%

$/M

Wh

(Nom

inal

Dol

lars

)

2007/082008/092009/102004/052005/06

prices during the off-peak as mainland generation is exported into Tasmania. This has already been observed since May 2006 when Basslink commenced operation, especially with dry conditions in Tasmania requiring high levels of import.


To model the bidding behaviour in the NEM, PLEXOS uses a long-run marginal cost recovery method, where generation companies will bid into the market so that they recover their LRMC. MMA has benchmarked its PLEXOS database to 2004 and 2005 calendar year market outcomes using this algorithm to ensure that the bidding strategies employed produce price and dispatch outcomes commensurate with historical outcomes. The PLEXOS model suggests that historical bidding strategies may result in less price volatility over the next few years as new capacity enters the NEM.

The results used from PLEXOS are averaged across the Monte Carlo simulations. Each simulation represents a pattern of random events such as unplanned plant outages. Averaging across the results on a half-hour basis will also result in a price path that shows less price volatility.


4 CONCLUSIONS

The analysis of wholesale costs for the supply of the regulated load from July 2007 to June 2010 modelled at the net system load plus the controlled load, referred to in this report as the “net system load”7, has shown that the peak energy is slowly increasing and is expected to continue to do so with increased penetration of air-conditioning load.

The analysis showed that the heating and cooling degree day threshold temperatures are also changing and converging toward one another. The regression analysis suggested rates of change greater than 1.0o C/year shown but that was not credible as a linear trend to be extrapolated. To obtain more sensible results the rate of change was limited to 0.5o C/year. There is scope for further analysis to assess the trends in threshold temperature. An exponential trend to a new steady state temperature might be an alternative hypothesis that would be worth testing in future work.

The load factor of the net system load may continue to decline slowly with increasing peak demands at the summer afternoon and winter evening peaks.

The trend in the load shape parameters together with the 2005/06 weather was used to forecast a net system load shape that would be consistent with a pool simulation using the same load shape. Load shapes were developed for 90%, 50% and 10% probability of exceedance (POE) weather by amending the extreme days in the 2005/06 weather year to fit the required temperature profile as published by TransGrid for each of these conditions. By this means we have a net system load shape and a system load shape that are consistent and it is easier to assess volume and price risk for different hedging positions.

It was initially assumed that a prudent retailer would acquire a flat and peak swap position over a period of three years leading up to the financial year of trading. However, in view of the recent announcement of the phasing out of ETEF, the costing analysis assumed that contract positions are taken up from 2006/07 over less than three years where necessary based upon lead time.

It was assumed that a $300 cap would be purchased to cover the gap between the peak swap capacity and the 10% POE peak demand. This is efficient because the average volume of energy in this capacity band is very small.

Contract prices for flat and peak swaps and flat caps were obtained from d-cypha and NGeS sources for the years from 2007 to 2010 on 22 November 2006. These contract prices were used to estimate future financial year contract prices having regard to the historical volatility of contract prices. A final contract purchase price was estimated by allowing for 90% probability that the actual contract price would be less than the chosen value.


The wholesale cost of energy with full risk management is expected to be about $55 to $57/MWh and the long-run marginal cost of service to be from $56.50/MWh to $62/MWh over the three year period. The costs of emission abatement and NEM fess will add an extra $5 to $6/MWh to these costs to give a range from $62/MWh in 2007/08 to $63/MWh in 2009/10. An annual summary of the results is provided in Table 4-1.

Table 4-1 Summary of Prices

Financial Year 2008 2009 2010Ending June



REC Cost $/MWh $0.88 $1.14 $1.43

NGAC Cost $/MWh $3.86 $4.12 $4.16

NEM Fees $/MWh $0.35 $0.33 $0.32

Ancillary Services $/MWh $0.30 $0.30 $0.30

Aditional Costs $/MWh $5.39 $5.89 $6.21

Total Purchase Price at NSW RRN $/MWh $60.79 $62.81 $63.26

ETEF Share 100% 77% 37%Note: Reflects costs at NSW RRN, Excludes Distribution and Transmission Losses

The analysis has demonstrated the changes in the patterns of loading and the uncertainty in the key cost components. It will be important for IPART in setting a maximum tariff to ensure that the trading range allows for these uncertainties and minimises the risk of market failure by not setting the maximum tariff for regulated customers too low. Each prudent retailer would need to operate on the same basis and would be prepared to pay a premium to ensure that its wholesale cost risk is properly managed relative to its retail obligations. Since each retailer would face the same risks each would be expected to pay similar premium levels to manage the wholesale price risk.

The wholesale cost analysis should be updated with the latest contract prices at the time that the retail prices are to be finalised.


APPENDIX A RENEWABLE ENERGY CERTIFICATES MODELLING

A.1 BASIS OF REC PRICE FORECASTS Under the MRET scheme wholesale market customers are required to purchase RECs equivalent to their liabilities under the scheme. The price of the certificates is primarily a function of the cost of supply of renewable generation, the actual level of the generation required to meet the renewable energy target and the structure of the wholesale market and the market for certificates. In this section, we describe the methodology employed to project renewable energy certificate prices and the key underlying assumptions.

The price of renewable energy certificates is affected by a number of factors:

• The nature, cost and available resource of renewable energy.

• Prices received for renewable energy generation in wholesale electricity markets.

• Revenue earned from other potential services provided by renewable generation, such as the ancillary services, avoidance of network costs, avoidance of waste disposal costs and green premiums.

• Short-term factors, such as variation in climate from year to year.

Some of these issues will be discussed in more detail below.

A.1.1 The Renewable Energy Resource

Renewable energy technologies are generally characterised by a number of features that will ultimately impact on the price of the certificates. Apart from the capital and operating costs, other factors affecting the choice of renewable generation options and therefore the price of certificates include:

• Constraints on fuel resource availability. This particularly impacts on the costs of biomass options, which may need guarantees of long-term fuel supplies. It also affects intermittent generation options, particularly the reliability of supply of the fuel (e.g., wind regimes and solar insolation levels).

• Changes over time in the capital costs of renewable energy technologies. The trend has been for declining costs of renewable energy capital costs as a result of technological enhancements and increasing scale of production.

• Lag times in developing renewable generation projects (including the time required to obtain approvals).


• Community concerns over the visual amenity or other pollution issues associated with renewable generation.

• Strategic factors that may cause investments in options that are not the least cost options.

Because of the dearth of site-specific information on renewable energy options, some retailers may contract with options with higher cost than would have been chosen on the basis of least cost for the system as a whole. A competitive market for renewable energy with well-informed participants would result in choices converging to least cost outcomes. Prices of certificates would be bound by the entry cost of the next highest cost option required to meet the target. Retailers who contract with higher cost options will face the risk of earning lower profits on their sales of electricity.

A.1.2 Electricity Prices

Output from renewable generation will either be sold on wholesale markets or will displace purchases from the wholesale market by end-use customers. Thus, renewable generators will receive revenue from electricity sales to wholesale customers.

The value of output for the renewable energy generators will be equal to the prices received in the pool market minus a loss factor covering losses in transmitting the electricity from the generator to the market. In some cases, renewable generators may confer an advantage to customers in lowering the network losses. The renewable generator could also capture part of the value of reduced losses.

Due to the operation of the NEM, the price of electricity varies significantly throughout the day. The highest prices occur at periods of high demand, primarily the morning and evening peaks, and low prices occur overnight as demand reduces. This diurnal cycle of wholesale prices has a large impact on the sales revenue earned by a renewable generator and the certificate price required to support the projects. Some renewable generators may have higher levels of generation during peak periods resulting in a higher average price for sales than a simple daily average. On the other hand, some other renewable generators such as solar hot water systems commonly replace off-peak electric systems, resulting in these assets receiving a much lower average value from the electricity price.

Additional Benefits to Generators

Some renewable generator options, particularly embedded and distributed generators, can provide other market services. Examples include ancillary network services such as voltage control. Other benefits include avoided network costs, lower losses, provision of steam from renewable based cogeneration and provision of other products or services (such as waste management).


In the NEM, the process for valuing some of these network services is under review. Of course, intermittent renewable generation options will be less successful in obtaining such benefits.

To the extent that renewable generation may confer additional benefits to electricity customers, the value of these benefits will impact on the REC price outcomes assuming a competitive market. The value of these services should fall in the range between the marginal cost of providing the service through renewable generation and cost of the alternative option for providing similar services.

A.1.3 Method

Projecting renewable energy certificate prices and the technology mix likely under MRET requires the use of a sophisticated model of the Australian electricity system. Our approach is to account for the interrelationships between the wholesale electricity market and the renewable energy market over the study period. Future REC prices are dependent on wholesale electricity market prices and the cost of renewable generation. In turn, the entry into the market of additional renewable generation will impact on wholesale electricity prices.

Geographical differences are also considered. Wholesale electricity market prices may vary by location, depending on local supply and demand factors and limits on transmission capacity. A region may have the potential for a large amount of renewable generation, but this potential may be thwarted by the lack of demand for electricity nearby. For the same technology, the costs also vary by location due to differences in fuel costs and transmission upgrade costs. Other benefits of renewable generation also vary by location.

MMA’s REC market model is based on the premise that a renewable energy certificate will trade at a value that will enable the marginal generator to operate economically, while meeting the mandatory interim targets. The value of a certificate may be determined from the difference between the levelised cost of generation of the marginal renewable generation unit and the electricity price obtained in the market for the thermal generation it displaces. Thus, the basis of the projections of the price of renewable energy certificates is that the certificate price will relate directly to the cost of renewable electricity generation. The renewable certificate will equate to the difference between the cost of the lowest cost renewable energy required to meet the mandatory target and the price for the electricity that can be obtained in the wholesale market. The cost of the last renewable option dispatched to meet each of the interim targets sets the market clearing price and the certificate price.

The prices forecast with this method represent an average price for contracted and large volume spot sales of RECs. Most RECs will be sold under bilateral contracts, with up to 20% of sales traded on the spot market.


An overview of the modelling process is shown in Figure A- 1. The approach is iterative since the timing and selection of renewable generation impacts on wholesale market prices and vice versa. The electricity prices that are produced as an output of the wholesale electricity market simulation model, Strategist, are used as inputs into the REC model. After running the REC model, any changes to the renewable generation options selected are input back into Strategist. The process may be repeated if it is deemed that substantial changes in REC price and technology mix are possible.

Figure A- 1 Overview of Method for Projecting REC Prices and Technology Mix

Solve for Prices in

Wholesale Electricity Models

Solve for REC Price

and Technology

Mix

Electricity Market Prices and

Generation by Unit

Identify Changes to Entry Timing based on LRMC for Conventional

Generation

Calculate Revenue Streams by Renewable

Technology and Location

Adjust Timing of New Entry

and Mothballing

When Stable, Capture REC

Prices, Technology Mix, GHG

Abatement

Calculate Levelised Costs

for Each Renewable Generation

Option

Calculate Levelised Costs

for Each Renewable Generation

Option

Solve for REC Price

and Technology

Mix

Calculate Revenue Streams by Renewable

Technology and Location

Solve for Prices in

Wholesale Electricity Models

Electricity Market Prices and

Generation by Unit

Adjust Timing of New Entry

and Mothballing

Identify Changes to Entry Timing based on LRMC for Conventional

Generation

When Stable, Capture REC

Prices, Technology Mix, GHG

Abatement

The estimation of REC prices is based on the assumption that the REC price provides the revenue, in addition to the electricity price, that is required to make the last required (marginal) renewable energy generator to meet the REC target viable. This takes into account an acceptable commercial rate of return to the project developer.

In a simple system the REC price would be determined by identifying the marginal generator and performing a simple subtraction of these two values. However, the following complications arise:


• Introduction of new renewable generators impacts on the electricity price paths, which may require iteration of the market price forecast and the REC estimation.

• The allowance of banking in the REC market results in the requirement for an inter-temporal optimisation. Under MRET, more RECs can be created in a year than required to meet the target to be banked and surrendered at a later date. This makes economic sense if the cost of creating the REC earlier than required is lower than the projected cost of purchasing a REC at a future date. The potential for banking means that the demand for renewable generation can be higher than the interim targets in the earlier years and lower than the target in the latter years. The effect of banking in terms of REC prices will depend on the level of banking and the costs avoided from creating surplus RECs.

• Currently installed and committed generators remain/enter the market regardless of the estimated economics. Because capital costs are sunk, these plants are assumed to be operating with just the marginal cost of generation considered in the modelling. Typically, these marginal costs are lower than the levelised costs for new units, so that committed plant are not likely to set the price in RECs in any year8.

• Generation resulting from the upgrade of large hydro units is treated in our hydro dispatch model to account for the additional dispatch that could be achieved with refurbishment to achieve higher efficiency in generation. This means that the additional capacity is treated as new generation capacity in the model, with full accounting of all costs incurred in the upgrade.

• Resource and other constraints limit the uptake of renewable generation. Resource constraints, for example fuel availability, are modelled by increasing the marginal cost of the resource.

The certificate price path is set by the net cost of the marginal generators, which enable the above conditions to be met and result in positive returns to the investments in each of the projects.

MMA has a detailed database of renewable energy projects covering existing, committed and proposed projects that supports our modelling of the REC price path. The database includes estimates of capital costs, likely reductions in capital costs over time, operating and fuel costs, connection costs, and other variable costs for individual projects that are operating, committed or planned9.

For this assignment, the data base was updated and revised. Currently, the data base comprises:

8 The marginal cost of an existing plant typically comprises only fuel and non-fuel operating costs (capital costs are sunk).

For new plant that is not as yet in the market, the marginal cost includes the cost of capital because the plant would need to recover capital costs to enter the REC market.

9 Committed plant means projects that are either under construction or have achieved financial closure. Planned projects are those being actively investigated.


• 466 eligible renewable generators, either existing, committed or planned.

• Existing RE generation accounts for 3,536 GWh per annum of eligible (above baseline) REC creation (excluding the proportion of generation sold on Green Power markets).

• Committed projects account for a further 3,792 GWh eligible REC creation, including 807 GWh attributable to solar hot water heater sales, with most of this generation coming into the market over the next two years.

• Planned or generic projects, excluding additional solar hot water sales, amounting to 36,323 GWh of eligible renewable generation.

Project costs have been obtained from published estimates of costs (usually capital costs) plus estimates of costs inferred from equipment suppliers, market data (for biomass fuel costs) and reports to Government. The costs are believed to be accurate to +/- 10% for existing and committed projects and +/- 20% for planned projects.

MMA has also developed a separate model for forecasting REC creation from SHW heaters taking into account the impact of a range of new State Government policies.

The MMA REC Model determines the future price path of RECs in the following steps:

• The costs of a range of renewable energy generation options have been determined as the levelised cost of generation using a 10.0%10 real pre-tax weighted average cost of capital over at most a 20-year investment horizon. The model considers the time from the commencement of generation to the end of 2020 for REC revenue but only considers energy (electricity) revenue beyond 2020 earned by the REC project if the 20 year investment horizon goes beyond 2020. The weighted average cost of capital estimate is also based on existing market rates for generation investments. Where data has been available the costs include the costs of connection to the grid, which can form a significant proportion of the capital costs of a project, particularly where no local transmission wires are available.

• The spot market price or wholesale electricity cost in each of the regions of NEM, SWIS or the Darwin-Katherine Grid has been used as the price that a generator could obtain for the power generated in the market. Wholesale electricity prices are determined on an hourly basis for each week of the study period, using a Strategist model.

• Assign regional wholesale electricity prices to all renewable projects in the data base according to location and start date. Weight wholesale electricity prices according to the generation profile of the renewable technology. For example, waste process generation would operate 24 hours per day and would therefore be represented by the average time-weighted pool price. Whereas, photo-voltaics would only operate

10 Based on debt to equity ratio of 75:25, real pre-tax interest on debt of 6.3% (9.0% in nominal terms) and real pre-tax

return to equity of 17%. A premium of 1% applies to biomass projects to account for fuel supply risk.


through daylight hours, achieving the prevailing market price for these hours only. Solar hot water systems although using solar energy during daylight hours, actually replace off-peak electricity usage so the surrogate price for this option is the off-peak price for the replaced energy.

• For each project, estimate any revenue from other sources such as fees for avoided landfill charges.

• Potential revenues from wholesale market transactions and other sources for each project are levelised for the life of the project.

• Subtract levelised revenue from corresponding renewable project levelised cost and then determine the merit order of the projects by ascending net costs (apart from those generators flagged as committed). The generation meeting the interim targets plus demand for banked credits in each year will determine which projects in the merit order will come on-line in a particular year.

• The generation output from each project is calculated from the MW and capacity factor for each project.

• For each selected new project the REC values over the remaining term of the MRET Scheme are discounted with the electricity sales income, and revenues from any other programs (e.g. steam sales). The discounted cash flow compared with the levelised cost indicates whether a given REC price path will justify the construction of a project.

• The REC path is optimised over the years of the program subject to the constraints indicated above.

• The plant installed in each year is determined by the economic viability subject to the REC price path, REC creation and surrender constraints.

• The resulting MW installed and generation levels are then input into wholesale electricity market model to determine the resultant pool price changes that in turn impact the REC prices.

• The process may be repeated until stable outcomes result.

Certificates are valid for all periods up to 2020. In this analysis banking of certificates over periods is allowed to occur where economic. This allows generators to hold their certificates until a later date when a more attractive price may be available. Banking of certificates may also reduce the total cost of the scheme by delaying the introduction of more expensive generation. It also means that all targets could be met by a group of renewable generators creating less than overall target for a period beyond 2010.

A.1.4 Assumptions


The price of certificates may be affected by:

• Regulations affecting supply, which will impact on the level and cost of each renewable generation technology. The Act defines eligible sources of renewable generation and defines restrictions on fuel sources, such as waste wood derived from native forests and plantations. Only renewable resources currently eligible are modelled.

• Other regulations that impact on the availability of resources, such as environmental and heritage regulations which may affect the amount of renewable generation occurring in some locations. The restrictions include: a ban on generation options close to urban areas, restricting the level of wind generation by location and restrictions on availability of fuels for biomass projects.

• The underlying cost of renewable energy technologies, including the cost of any network upgrade required to supply the grid and ancillary services. Network upgrade costs are included in the modelling where information is available. An assumption is also made for a cost impost on intermittent generation alternatives (primarily wind generation) for an additional cost for the provision of ancillary services11. This is assumed to be about $5/kW.

• Prices received for renewable energy generation in wholesale electricity markets. Prices received are affected by a number of factors, including the reliability of generation and the location of the generator. For intermittent generation, retailers may wish to discount the prices received for intermittent generation in contracts to compensate for the non-firm nature of generation. In our model, we use average hourly prices to determine revenues earned for intermittent renewable generation, but these are discounted by 10% to reflect the uncertainty due to intermittency of the generation.

• Revenue earned from other potential services provided by renewable generation, such as the ancillary services, avoidance of network costs, and avoidance of waste disposal costs. In the modelling, revenue from other sources is assumed to be zero.

Because of banking, current prices in the RECs market will be based on the expectations of future market conditions of all traders involved. Thus, the current price will be an expected price based on a number of possible future market scenarios and the probability of these scenarios eventuating. Other short-term factors may also impact on the price.

In our modelling, we attempt to project certificate prices for a most likely outcome in terms of electricity price, availability of renewable resources and generation costs.

11 Because generation from a wind farm can vary from minute to minute, additional resources are required to stabilise

voltage on associated network elements. See Arnott, I. (2002), “Intermittent Generation in the National Electricity Market”, National Electricity Market Management Company, Melbourne


Therefore, we have not explicitly modelled the impact of short-term or other factors that may affect expected prices.

Costs of renewable generation

There are several sources of supply uncertainty that could affect the forecasts of REC prices. Generation from some renewable energy options is intermittent. This affects the reliability of supply and the prices received for the energy. Depending on the penalty for non-compliance, the unreliability of supply may also lead to a high level of renewables being required in order to guarantee the targets are achieved. Risk averse retailers may over-contract in order to ensure they can meet their targets taking into account the probability that the renewable generator may not generate the contracted quantity due to adverse climatic conditions. Or they may contract for that generation at a discount, with current indications being that discounts of up to 10% on expected spot prices are being sought by retailers.

Data on the level of variability of renewable options are sparse. The two most affected technologies are wind and hydro-electric generation. Preliminary data on wind generation indicates a year on year variability of plus or minus 10 per cent (95% confidence interval). Variability in annual hydro generation is about plus or minus 11 per cent based on data from the Snowy Mountain Scheme and Hydro Tasmania, although generation from these schemes has been lower than the long-term average over the past three years.

However, the impact of intermittent supplies on renewable certificate prices is likely to be minimal. The reasons for this include:

• Retailers can use the banking provisions of the scheme to bank some of the certificates in years when renewable energy generation is higher than expected for use in years when generation is lower than expected.

• Potential cross-correlation in the supply of renewable energy resources by type and location of the resources. Low wind generation in one region may be made up for by higher than average wind generation in another region or by higher than average generation by mini-hydro options. There is a dearth of data on the potential for cross-correlation in renewable energy supplies.

• Usage of biomass or co-firing options, which have more stable supply.

Another source of supply uncertainty is the potential limit on the availability of renewable energy resources due to economic or technical circumstances. For example, some renewable energy resources are only available for limited periods during the year. Bagasse is only available during the sugar cane harvesting period of May to November. The unit cost of the renewable energy is increased not only because of the lower level of utilisation of assets, but also because the outputs are typically sold in the lower price periods in the electricity market (from May to November of each year). Storage facilities to


enable year round usage of bagasse would add to the cost of bagasse based generation. The additional cost of this storage has been included in the analysis.

Because many of the biomass fuels are by-products of other productive activities, their availability is subject to economic factors affecting those activities. For example, bagasse is a by-product of sugar cane production and the amount of sugar cane crushed. Supply of sugar cane is variable due to the variability of sugar prices on world markets and variable weather conditions (which can also affect fibre content). This is included as a premium on the discount rate of 1% to reflect this uncertainty.

The future costs of renewable projects also depends on the forecast reductions in capital prices resulting from technological improvements, the value of the relevant exchange rate and the ability of the project to obtain additional government support. In recent times, increases in labour and material costs have boosted the capital cost of both renewable and fossil fuel generation options. Based on anecdotal evidence plus the cost of recently announced projects (e.g. Hallett Wind Farm) the capital cost of all proposed renewable generation projects have been increased by 10%.

Obviously changes to these costs from those assumed would have a significant impact on prices. Higher capital costs would impact on prices, particularly in the latter period of the scheme when high capital cost options are setting the certificate prices. Increase in fuel costs will also have a moderate impact on prices. This is because such cost increases would increase the cost of biomass generation options as well as change the profile of generation to higher cost options such as wind generation.

Assumptions on the cost of renewable generation are shown in Table A- 1. The net cost curves12 for available renewable energy in Australia are shown in Figure A- 2. This does not include the creation of RECs from additional SHW heater sales, and which can amount to up to 800 GWh by 2010 for REC prices equal to or greater than $35/MWh.

Table A- 1 Long run average costs of renewable generation options in 2007, $/MWh

Renewable Generation Type Minimum Maximum Hydro-electric 78 276 Wind 70 120 Biomass 69 158 Note: Long run average costs represent average cost (including capital, transmission, operating and fuel costs) calculated using 10.0% pre tax cost of capital. Costs are in mid 2005 dollar terms.

12 Net cost is the long run average cost after deducting for revenue earnt on electricity markets.


Figure A- 2 2007 Net cost curves for renewable generation, mid 2005 dollar terms13

0

50

100

150

200

250

300

0 5000 10000 15000 20000 25000 30000 35000 40000 45000

Eligible generation, GWh

Long

run

mar

gina

l cos

t, $/

MW

h

A.1.5 Scenarios

Many factors will impact on pool prices and the outcomes in the market are uncertain. To reflect the uncertainty, high and low REC price scenarios were also developed. Assumptions underpinning the base, high and low scenarios are shown below.

Table A- 2 Assumptions for base , high and low REC price scenarios:

Variable Base High Low

Fuel costs for renewables

Base biomass prices 10% higher biomass prices

10% lower biomass prices

Capital costs of renewable generation

Base capital costs (approx 2% pa reduction)

1% pa reduction in real capital costs

3% pa reduction in real capital costs

Limit on solar hot water heaters

1,600 GWh 1,100 GWh 2,500 GWh

Resource limits None Low cost wind generation reduced by 10% to reflect planning restrictions

None

*REC prices are calculated from the difference between the levelised cost of the marginal renewable generation unit and the electricity price obtained in the market for the thermal generation it displaces. This means that low electricity prices will lead to high REC prices and vice versa.

13 Covers existing, committed and proposed renewable generation projects


A.1.6 Short term outlook

REC prices have been falling since the start of the year. At the beginning of 2005, the REC spot price was around $38/MWh, whilst the forward price for 2008 was around $46/MWh to $48/MWh. Since then, the price has fallen, at first modestly, but over mid 2005 sharply. Spot prices fell to around $27/MWh, whilst the forward price for 2008 fell to $34/MWh in May 2004. For the remainder of 2005, prices steadied at those levels.

Since the start of 2006, however, prices have again fallen to around $23/MWh for 2006 and $31/MWh for 2009. The latest trading data from NGeS in Table A- 3 shows that prices have fallen even further to $18/REC. NGeS states that there is very light volume at these prices as the REC market awaits further developments. The implementation of the Victorian renewable energy target would be expected to divert renewable resources away from MRET toward the Victorian target. This would be expected to support REC prices.

Table A- 3 REC Trading data for August 2006

Source: NGeS “Green Room” 28th August 2006

The fall in spot prices has been due to a perception of a large surplus of banked RECs and solar hot water heater operators wanting to liquidate their stock of RECs. Liable parties are holding back their purchases of RECs until they perceive the price of RECs has bottomed. Liable parties have been reluctant to enter into long-term contracts with renewable generators until they get a clear picture of the supply/demand balance for RECs.

MMA has developed a database which compiles statistics on the number of RECs created and information on all renewable generators (including proposed projects).

From this data base, several interesting statistics emerge that help to explain the recent fall in prices. Some of the key points are as follows:

• MMA calculations on data in the REC Registry reveals that at the end of 2004, the level of excess certificates created (that is, the level of REC creation to the end of 2004 above the level of liabilities to surrender RECs to the end of 2004) was 6.0 million (i.e. 6,000 GWh). Other sources (see NGeS) put the surplus at 5.2 million (i.e. 5,200 GWh). These estimates are equivalent to between 473 GWh per annum to 546 GWh per annum, reducing the target to be met to around 9,000 GWh from 2010 onwards.


• MMA estimates reveal that generation from existing RE generators and committed (i.e. projects under construction or financially closed) is about 7,327 GWh per annum. MMA calculates that by the end of 2007, there will be a further surplus of 3,421 GWh of RECs on top of the surplus of 6,000 GWh already created.

• Our latest modelling reveals that the total generation from 2010 onwards is likely to peak at around 8,400 GWh not the 9,500 GWH target level (the actual level of generation is lower due to the surplus of RECs created). This leaves a deficit for future generation of around 1,073 GWh (i.e. 8,400 – 7,327 = 1,073). Thus new RE generation and solar hot water heaters will compete for this market.

• The solar hot water heater market is expected to grow at 10% per annum until 2010, allowing an additional 790 GWh of RECs to be created each year from this source (on top of the 800 GWh per annum in 2004). Deducting this from the total pool of RECS available for new projects means that new RE generators will be competing for 283 GWh per annum of RECs. This is equivalent to around 100 MW of additional wind generation or 40 MW of biomass or geothermal generation operating at base load.

There is currently 36,000GWh per annum of planned RE capacity competing for this 283 GWh per annum of spare REC demand. Hence, the downward pressure on prices.

Obviously, many of the planned projects will not proceed. However, the market still needs to achieve a REC price that recovers the cost of RE generation not recovered from sales on the wholesale electricity market.

Our data base indicates the net levelised cost (that is the long run average cost of new renewable generation minus the average electricity price likely to be received) for the most likely RE projects to proceed is likely to be around $33/MWh to $40/MWh range. This should set the floor price for RECs under long-term contracts.

In the short term, prices are likely to be determined by the level of price required to make a solar hot water economic relative to conventional hot water heaters. MMA analysis indicates this to be around $23/MWh to $27/MWh, below which it would be uneconomic to purchase a solar hot water heater. Interestingly, the current spot price has until recently hovered around this level and then fallen to $18/MWh. We do not expect this price to be sustainable, especially since the low rainfall is going to reduce the REC supply from existing hydro resources.

Thus, our current view is that prices have bottomed and that they are likely to start rising again once retailers enter back into the market for long-term contracts. However, this outlook depends critically (as discussed below) on two factors:

• The growth in RECs supplied by solar hot water heaters. There is a reasonable probability that there could be significantly more RECs created from solar hot water heaters than assumed in our model, leading to prices remaining at current levels.


• The amount of above baseline generation from RE generators in operation before 1997. Estimates of above baseline generation varies widely (we assume about 1,600 GWh). Higher levels than assumed would put downward pressure on REC prices.

A.1.7 Long-term outlook

Forecasts of REC certificate prices for the median scenario, in mid 2005 dollar terms, for the current MRET targets are shown in Table A- 4. REC prices are forecast to increase slightly from current level of about $18/MWh to just under $40/MWh by 2008. This increase is due to the higher cost of bringing new plant on during this period as lower cost options have already been commissioned. Constraints in resource availability also limit the amount of some low cost options entering the market. Further, costs increase due to the shorter period available under the MRET Scheme to recover revenue through the sale of RECs.

This long-term outlook takes into account the fact that existing hydro-electric generators can generate significant amount of RECs each year, assuming adequate water inflow into their dams.

Table A- 4 REC Price Forecasts – Median, High and Low Cases, $/MWh, Real mid 2006 Dollar Terms

CY Ending Med High Low Penalty 2006 28.0 29.0 26.0 57.1 2007 36.9 49.3 26.0 55.7 2008 39.4 51.2 26.0 54.4 2009 39.4 53.1 26.0 53.1 2010 39.4 51.8 26.0 51.8 2011 39.4 50.5 26.0 50.5 2012 39.4 49.3 26.0 49.3 2013 39.4 48.1 26.0 48.1 2014 39.4 46.9 26.0 46.9 2015 39.4 45.8 26.0 45.8 2016 39.4 44.6 26.0 44.6 2017 39.4 43.6 26.0 43.6 2018 39.4 42.5 26.0 42.5 2019 39.4 41.5 26.0 41.5 2020 39.4 40.4 26.0 40.4

Note: Prices apply to those available for long-term contracts for large parcels of RECs. Source: MMA analysis

Prices are forecast to be subdued in the short term as competition has increased in the REC market reflecting a surplus of generation projects competing to fill a shrinking market. Because of this surplus, retailers are reluctant to enter into long-term contracts with RE project developers until they see the market settle. This means that the lowest cost projects only will get up. It also means that price support strategies adopted by many of the larger holders of certificates are no longer working to support prices at higher levels.


Entry of new renewable generation, mainly solar hot water heaters, creates further competitive pressure dampening prices of certificates. Under the modelling, about 1,600, 000 RECs per annum are created from new solar hot water after 201014.

These pressures mean the prices for contracted RECs are forecast to rise only steadily from current levels of around $27/MWh in 2005 to around $40/MWh in 2008. From 2008, REC prices remain steady at just under $40/MWh. By then, enough renewable generation has entered the market and prices are shadowing the cost of the next increment of REC capacity that would be required to enter the market.

Prices for the high and low scenarios reflect variations in electricity demand (lower and higher than the base case respectively) and cost of renewable generation (higher and lower than the base case respectively).

Please note that the price forecasts reflect the prices for long-term contracts needed to support the level of renewable generation required to meet the MRET target. The low to high range in the forecast prices does not reflect short-term factors such as variations in annual generation from intermittent plant such as hydro-electric generation, higher or lower level of sales of solar hot water and higher or lower electricity prices in any one year. Short-term factors can add about $10/MWh or subtract about $15/MWh to the forecast prices for the base scenario. However, our analysis indicates that prices are unlikely to fall below $25/MWh for sustained periods as prices below this level start to make the solar hot water heater sales unattractive (that is, the price of RECs below this level is insufficient to make the purchase cost of a SHW comparable to conventional hot water heaters for a growing proportion of the replacement heater and new heater market).

The forecasts for the Base, High and Low cases are shown in Figure A- 3. The forecasts are underpinned by the costs for renewable generation. Over time, there are two trends in the assumed cost data:

• Costs increase with year. This is because of the assumption that the scheme expires in 2020, so that the later a plant enters the market the less time it has to recover costs from generation of RECs. This applies to all renewable generators apart from solar hot water heaters (which are not affected due to the special deeming provisions which enable them to earn 10 years worth of RECs upon installation15).

• The increase in costs is partly mitigated by the increase in availability of new renewable generation options.

• The increase in costs is also partly mitigated by the fact that a renewable project could continue to earn revenue beyond 2020 from sales of electricity and perhaps sales of carbon credits beyond 2020 (the latter is not assumed in this analysis).

14 Note that a new solar hot water heater is given 10 years worth of deemed RECs on installation. Thus, the purchase of

solar hot water heater is not impacted by developments post 2020. This is why the model continues choosing solar hot water heaters beyond 2010.

15 Note, however, that ORER has flagged that the deemin period for solar hot water heaters may be reduced psot 2010.


The Low Case indicates a price staying steady at $26/MWh. This reflects the key assumption that certificate creation from solar hot water heaters reaches a level of 2.5 million per annum. With recent announcements of new wind and biomass projects achieving financial close, creation of RECs from solar hot water heaters at this level would mean there would be no further need for new renewable generation plant, so that the price would be at the floor price for attracting solar hot water heaters into the market. We assume this floor price at $26/MWh.

Figure A- 3 Base, high and low REC price projections, real mid 2005 dollar terms

0

10

20

30

40

50

60

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

Calendar year

$/M

Wh

Med High Low Penalty

A.1.8 Other risks

The views expressed above are dependent on a number of assumptions. The key assumptions underpinning this outlook are:

• Uptake of solar hot water heaters. With changes to State government policies supporting solar hot water heaters, the uptake of solar hot water heaters may be greater than anticipated above. Some commentators believe the additional contribution of solar hot water heaters could be as high as 2,500 GWh of REC compared with the 1,600 GWh assumed in the base case.

• Level of surplus RECs at the end of 2004. MMA’s calculation of this surplus is based on created RECs (that is registered with the Office of Renewable Energy Regulator). Some participants may be withholding potential RECs by generating but not yet registering the RECs. Thus, the level of surplus RECs may be significantly above 6,000


GWh estimated by MMA. Again, higher levels of surplus RECS at the end of 2004 would put downward pressure on long-term prices.

• On the other hand, the capital costs for new (planned) renewable projects may be underestimated. The capital costs for new projects are based on published data for projects under construction or recently completed. Since these projects were completed, capital costs have increased due to higher material and labour costs. We have allowed for a 20% once off increase in capital costs, but the actual level may be a lot higher than this. For example, new conditions imposed by NEMMCO on wind (or other intermittent) generators may impose additional costs on some wind farm projects. If this is the case, then long-term prices could be higher than we have estimated.

Another key uncertainty is whether or not the changes to the operating rules of the scheme as recommended by the MRET Review Panel are adopted. They are likely to be adopted but recent MMA analysis indicates the impact will be minimal as the changes are likely to have a minimal impact on the choice of new RE generators into the market.

Retailers and renewable energy generators face a number of risks in trading in the certificate markets. The risks include:

• The possibility that renewable energy costs could fall over time, which would discourage risk averse retailers from contracting fully to meet their future renewable energy targets. Contracting fully with existing suppliers could expose the retailer to the risk that certificate prices could be lower than expected in the future making it difficult to pass these costs on to contestable customers. One method of avoiding this risk is to strike contracts with a higher initial price and declining prices over time. This has become more common in recent years.

• The intermittent nature of generation for some renewable energy options. In each year, more or less renewable energy will be generated depending on weather conditions or fuel availability. Banking of certificates would tend to reduce this risk.

• The possibility that electricity demand and therefore the electricity prices and certificate prices could be higher or lower than projected.

• Political risks associated with the possibility that future governments may alter the operation of the scheme or increase or reduce the targets. The risk of reducing targets would be appear to be small at this stage. As argued elsewhere, there is little chance of targets increasing under the current Federal Government.

These risks act on market prices for RECs. MMA analysis indicates significant upside and downside risks in spot prices, particularly due to short-term factors. The analysis indicates that the spot price could vary by as much as +$10/MWh to -$15/MWh due to short-term factors alone.


In summary, REC prices are likely to remain around the $28/MWh to $40/MWh level in real terms over the long-term under base case assumptions. The analysis indicates subdued prices in the near term as more renewable projects are commissioned. But there is a slight rise in the longer term as more expensive projects are required to meet the higher target and as the current overhang of projects dissipates. The prognosis of prices increasing back to levels around $39-$40/MWh depends crucially on a small rate of decline in capital costs of renewable technologies and rational decision making so there is not a surplus of renewable generation to meet the scheme.

The key uncertainty is the availability of RECs from solar hot water sales. In the base case, the projection is RECs created from solar hot water sales to reach around 1,600 GWh by 2010. This represents a substantial increase from the current levels of around 939 GWh (as occurred in 2005) and is a key factor putting downward pressure on REC prices over the next few years.

The increase in sale of solar hot water heaters is driven by a range of new programs implemented by State Governments. For example, the Victorian Government has mandated that new homes must either install a solar hot water heater or a rainwater tank to obtain development approvals. The Queensland Government is also implementing a similar scheme. An Attractive REC price is also driving growth, with a price of around $26/MWh being required to make a solar hot water heater economic relative to alternative solar hot waters in the replacement market in the main States of Queensland and NSW.

Nonetheless, there is a high level of uncertainty over the total level of RECs created from solar hot water heaters. Restricting the maximum sales to just 1,100 GWh per annum (instead of 1,600 GWh per annum) would see prices increase in both the short and long-term, potentially reaching the penalty rate by around 2009. Conversely, increasing the number of RECs created to 2,500 GWh would see prices drop to around $26/MWh in the long run. Thus, developments in the SHW market need to be monitored closely.

Another key uncertainty is the electricity price received by renewable projects. Long term uncertainties in the electricity price are reflected in the high and low price forecasts shown above.

A.2 Comparison with forward prices MMA forecast of REC prices are based on the fundamental drivers affecting supply and demand for RECs. Thus, the forecasts are based on the assumption that the market price is largely determined by these drivers and that the assumptions for the drivers are reasonably accurate. Thus, the forecasts should be interpreted as prices that would emerge if the market for RECs is highly competitive and that the cost drivers are within the range assumed for the modelling. The forecasts should also be interpreted as the prices that would apply to long-term contracts between RE generators and liable parties.

We compare MMA forecasts with forward prices currently being quoted in the market as shown in Figure A- 4. The forward prices are sourced from NGeS (Next Generation


Energy Solutions) represent actual forward prices as represented by the last trade for parcels of RECs of 10,000 or more. NGeS only provide forward prices to 2010.

MMA’s forecasts are noticeably higher than the forward prices quoted by traders for spot sales. Possible explanations for the difference include:

• Sentiment is depressed in the market due to the uncertainties over the supply/demand balance. Hence, the market is putting emphasis on range of short term factors.

Figure A- 4 Comparison of MMA base case forecast and market forward curve (June 2005 prices)

$-

$5

$10

$15

$20

$25

$30

$35

$40

$45

2006 2007 2008 2009 2010

Calendar

$/M

Wh

(mid

200

5 do

llars

)

MMA NGeS - spot trade

NGeS prices correspond with those reported in “The Green Room” on 5th June 2006.

• The forward market is illiquid and does not reflect the market prices for over the counter transactions. The forward prices are determined from a limited number of trades and may not reflect the prices for contracts, which comprise the bulk of the trades. In particular, it appears that forward prices are based on small parcels of RECs, which may reflect short term factors rather than longer term fundamentals. Anecdotal evidence suggest that contract prices for unstapled RECs (i.e. where the RECS are contracted for but not the electricity output of a renewable generator) are obtaining a premium of $2/MWh to $4/MWh above quoted forward prices. Furthermore, some retailers (e.g. AGL) have recently invested in new renewable generation projects at a cost of about $38/MWh suggesting they have a view that the REC price for long-term contracts will go towards that level.


APPENDIX B OPTIMISING CONTRACT POSITION TO MINIMISE COST

One of the challenges of retailing is to minimise a conservative value of wholesale energy cost allowing for contract prices and expected volatility of spot prices which affects that part of the retail load that cannot be reasonably hedged. This project was developed with a view to establishing a robust method for estimating an optimal contract position including flat swaps, peak period swaps and $300 caps applying in all hours that would minimise the cost that would be expected to be exceeded with only a 10% probability in each quarter. The method was tested using the 2005/06 regulated load and the actual NSW pool prices. The method took the view that a practical forward contract position would aim to fix a contract volume for the four quarters for each of these products and that the products would have been purchased well beforehand based on an expected volume. This analysis did not look at uncertainty in volume which could be accommodated by replacing swaps or cap positions with options for these products that could be exercised closer to the trading period.

B.1 Contract Level Relative to Average and Minimum Load Previous estimates of wholesale cost by Allen Consulting Group and Charles River Associates had been based on the principle that a prudent retailer would buy swaps to cover average demand and additional caps to cover expected peak demand under assumptions of medium growth and 50% POE peak demand1. In the Allen’s Report of November 2004, AGL is quoted as saying that it would cover minimum demand with swaps and the remainder with caps, so as to avoid over-contracting. The Charles River Report of December 20032 set the flat swap level at the minimum working day load which would be a little higher than the weekly minimum load. However there does not seem to be any particular justification for these estimates.

B.2 Effect of Contracting Levels on Price Over-contracting would have the effect of reducing pool prices so a realistic analysis should allow for the fact that as contract volumes rise the spot prices fall. This exacerbates the cost of over-contracting because additional difference payments are required to be paid by the retailer without commensurate value from having purchased the corresponding load for sale at a low price. Whether or not this in an important consideration for a retailer depends on whether its contracting behaviour is of significant magnitude to move the market. For a mass market new entrant retailer with relatively small trading volume, this contract volume effect would be minimal and accordingly we have conducted the following analysis on the assumption that spot prices are not affected by contracting levels. If spot prices were to rise as contracting levels fall, then the optimal

1 “Supplementary Report, Energy Wholesale Price Study” by The Allen consulting Group, 22 November 2004, page 12 2 “Electricity and Gas Standing Offers and Demmed Contracts (2004 -2007)”, Charles River Associates, December 2003


peak swap volume falls slightly in favour of higher cap volume if the product prices do not change with varying spot prices. This occurs because the over-contracted amounts are settled at lower spot prices, which is less favourable for the retailer. Such analysis is quite complex and considered not particularly helpful to the current problem.

B.3 Methodology MMA has explored the problem of estimating optimal contract levels and has confirmed that the optimal level is within 10% of the average load in the applicable trading period and certainly not the minimum load. We have taken the historical NSW net system load profile as provided by EnergyAustralia and processed it separately for each quarter from July 2005 to June 2006. The results for Q1 2006 are illustrated here.

The steps were:

1. Assume that the spot price is log normally distributed in each settlement period

2. For each half-hour calculate the rank of that load in the quarter where 0.0 represents minimum load and 1.0 represents maximum load. This is a more convenient form rather than relating the price profile directly to load level because high prices are expected at times of peak demand which is a relative effect unrelated to the load level itself over time. The general form of this function should change slowly over time as the patterns of system and net system load change.

3. Examine the relationship between NSW spot price and the load rank as shown in Figure B- 1. It is apparent that there is quite some price uncertainty at any given load level as would be expected. The scatter of this chart is somewhat less than if the price is plotted versus the load directly because the high but infrequent load levels are pushed toward the rank values greater than 0.9 even if the load is only 0.8 of the peak value.

4. Establish a relation ship between the Net System Load Rank and the average of the natural logarithm of the NSW price. This is equivalent to estimating perfect synchronisation of price and demand. This is shown in Figure B- 2 using the following equation (1) to describe the logarithm of the price (y) in terms of the load rank (x). The capital letters A to G are constants found by a solver method to minimise square errors of the price. The minimum function excludes values above the market cap price of $10,000/MWh. The constants are found separately for off-peak periods and peak periods. Figure B- 2 shows the curve fit for the off-peak periods which includes some high prices on weekends.

y = Min( log(10,000), Ax + B + (C/(D- x))^E + F exp( G x) ) (1)


Figure B- 1 Relationship Between NSW Spot Price and Net System Load Rank

NSW Price Versus Net System Load

2

3

3

4

4

5

5

6

6

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

Rank of Net System Load

LN (N

SW S

pot P

rice

$/M

Wh)

Figure B- 2 Fit of Average Relationship Between NSW Spot Price and Regulated Load Rank

NSW Price Versus Net System Load

2

3

3

4

4

5

5

6

6

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00


LN (N

SW S

pot P

rice

$/M

Wh)

5. From the residual errors, estimate the standard deviation of the logarithm of price as an exponential function of the regulated load rank was derived. This appears as a linear function on the logarithmic scale in Figure B- 3.


Figure B- 3 Variance of the Logarithm of NSW versus the Regulated Load Rank for Off-Peak Periods

Off-Peak Variance from Trend Versus Net System Load

0.000000

0.000000

0.000001

0.000010

0.000100

0.001000

0.010000

0.100000

1.000000

10.000000

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1


6. The residual values of the price deviations between settlement periods were then examined to assess the duration over which significant correlation of prices variation occurs. This was intended to support the quantification of cost uncertainty implied by a given hedge position. It was found that overnight, prices are insignificantly correlated between settlement periods more than 3 hours apart. A similar duration was found during daytime peak periods and on weekends. Simple models of the correlation were established by regression of the observed data using either exponential functions or quadratic functions. An example of the correlation for weekend prices is shown in Figure B- 4. The more important correlations were deemed to be those for settlement periods separated by less than 4 hours and by 24 hours plus or minus 3 hours. The positive correlations were applied when estimating the variance of total pool payments. The remaining negatively correlated periods were assumed to approximately cancel the positive correlations that were neglected over longer periods and between peak and off-peak periods. This was a necessary approximation to contain the magnitude of the analysis.

7. In addition, the variation in the average off-peak prices (and similarly the peak price) over a quarter were also allowed for based on historical price variation in the NSW pool. On a quarterly basis this would be about $2.20/MWh in off-peak periods for an average price of $20/MWh and about $10/MWh for a quarterly peak period price of $45/MWh. These values were derived according to the estimated average price using the functions shown in Figure B- 5. The orange


circles show the relationship for the last twelve months ending August 2006. The red line shows a quartic polynomial regression of the long-term relationship after adjusting prices to June 2006 dollars.

Figure B- 4 Correlation of price errors between settlement periods during a weekend off-peak period.

Correlation of weekend price variation

-0.600

-0.400

-0.200

0.000

0.200

0.400

0.600

0.800

1.000

1.200

0 2.5 5 7.5 10 12.5 15 17

.5 20 22.5 25 27

.5 30 32.5 35 37

.5 40 42.5 45

Hour separation

Cor

rela

tion

Coe

ffici

ent

Used these two positive parts of the correlation

Figure B- 5 Estimated Relationship between Annual Price Volatility in peak and Off-peak periods and the Expected Spot Price.

Volatility of Annual NSW Off-Peak Pool Price($Jun 2006)

$-

$0.50

$1.00

$1.50

$2.00

$2.50

$3.00

$3.50

$4.00

$4.50

$15 $20 $25 $30 $35 $40

SMP

Ann

ual V

olat

ility

$/M

Wh

ActualFit

Volatility of Annual NSW Peak Pool Price($Jun 2006)

$-

$5.00

$10.00

$15.00

$20.00

$25.00

$15 $25 $35 $45 $55 $65

SMP

Ann

ual V

olat

ility

$/M

Wh

ActualFit

8. Firstly for the off-peak period the total cost of the flat contract plus the expected spot trading in the off-peak period plus a 90% confidence margin for the uncertainty in spot revenue over the quarter was estimated using the above analysis. The flat contract volume was adjusted to minimise the cost assuming that any change in the flat contract volume would be offset by an equal and opposite change in the peak contract volume. This allows the off-peak trading risk to be


decoupled from the peak period trading risk in the process of estimating the flat contract requirement. The change in value for Q1 2006 is shown in Figure B- 6 as an example. As the contract level increases beyond the load level, the increasing difference payments become more and more unfavourable. For low contract levels the additional cost of the peak period contract cancels the savings from the low cost pool purchases in the off-peak period. In this case the contract level is 65 MW in excess of the average off-peak load, probably because of the substantial load and price risk on weekends when the load is much higher than the over-night load. The asymmetry of price risk makes a contract volume higher than the average load more beneficial. The cost analysis takes account of the saving on the peak contract volume.

Figure B- 6 Example of Basis for Optimal Flat Contract Volume (Q1 2006)

Cost for Flat Contract for Q1 2006

$82.00

$82.50

$83.00

$83.50

$84.00

$84.50

$85.00

$85.50

800 900 1000 1100 1200 1300 1400 1500 1600 1700

Flat Contract Volume

Average off-peak load

9. The same analysis was then conducted for the peak period with a view to estimating the peak contract volume that would minimise the cost of the peak contract plus the spot trading due to volume variations. It was assumed that a cap would be applied to make up the difference between the peak plus flat contract volume and the peak demand in the quarter. The total costs of the peak contract, the spot price exposure at a 90% coverage level and the cost of the cap less the corresponding difference payments was evaluated to determine an optimal combination of cap plus peak contract. The optimal value depends on the relative prices of these two products. The cost profile for variation in the peak contract volume is shown in Figure B- 7. Note that the optimal peak swap volume is almost twice the average peak load less the average off-peak load. This would reflect the assessed value of the swap relative to the cap for the regulated load profile. It is


apparent that using the average load would slightly over-estimate the hedged wholesale cost by about $0.24 M which corresponds to $0.08/MWh in this case. Hence errors in setting contract volumes are not necessarily critical but we have attempted to make an economic estimate.

Figure B- 7 Example of Basis for Optimal Peak Contract Volume (Q1 2006)

Cost versus Peak Contract Volume for Q1 2006

$37.50

$38.00

$38.50

$39.00

$39.50

$40.00

0 50 100 150 200 250 300 350 400 450

Flat Contract Volume

Average Peak Load Minus Average Off-peak

Load

One significant issue is the extent to which contracting strategy by a mass market new entry retailer can affect spot prices. We have confirmed using our models that they show that if contracting additional volume would depress spot prices then the optimal contract position is at a lower level. This occurs because the value of the subtracted portion of the contract volume is offset by the additional difference payments that are made to compensate for lower spot prices.

It is impossible to identify the extent to which spot prices would depend on contract volume in a generic way because the answer depends on the volume of contracts traded and whether the contracts are traded with generators or with other retailers. Therefore we have not attempted to take this effect into account in the present study. We have assumed that the mass market retailer’s contract trading is too small to influence the spot market prices.

B.4 Application to Modelled Spot Prices This method was used with the PLEXOS derived prices by the following method on the basis as described:

• Using the method described in Appendix D , the net system load was derived for corresponding peak system load levels at 90%, 50% and 10% POE so that we could


separately assess the effect of peak load variability. The cap volume was associated with the 50% POE peak demand but the volatility of the residual pool payments was included in assessing the full cost.

• For each corresponding system load profile a number of PLEXOS simulations were run as shown in Table B- 1 and the expected pool price was obtained for each half-hour. These average prices were not as volatile as the historical sample around the estimated average values. Therefore, it was not feasible to imply hour to hour volatility from these half-hourly average prices. We analysed the individual PLEXOS samples and we found that the residual variations had a similar correlation between time periods as observed from the 2005/06 sample. We therefore took the view that the time correlation could be applied to the price variations due to the inconsistent relationship with the net system load as well as the normal spot price variations from hour to hour.

• The results for the three sets of load profiles were weighted according to the ratios shown in Table B- 1 when calculating the optimal contract volumes. These weightings have been previously derived based upon the distribution of NSW summer peak demand at the three probability levels to approximate a continuous distribution of peak demand exposure by means of the three values according to a variation of the method presented by Miller and Rice.

Table B- 1 Weighting of scenarios to obtain expected value

Probability of Exceedance

90% 50% 10%

Weighting of scenarios for equivalent

26.4% 47% 26.6%

Number of PLEXOS simulations

20 30 50

• The contract costs were estimated assuming that an optimal contract volume is achieved by the time the load is delivered at the forward cost. The uncertainty in the sales volume affects how much of the volume is purchased under swap and cap options that remain flexible until they are called. A premium is added to the contract price proportional to the amount of the contract that is optional.

• The total costs were then added up as shown above in Table 3-1.


APPENDIX C COMBINING VOLATILITY PARAMETERS IN THE PLEXOS PRICE ANALYSIS

As discussed in Appendix B , a relationship between the Regulated Load Rank and the natural of the logarithm of the NSW price was established. The volatility of the NSW price can be modelled by the standard deviations of the logarithm of price derived from this function which assumes a log normal distribution of price in each settlement period. PLEXOS modelled spot price results were used as the NSW prices for 2008/09 and 2009/10 and the standard deviations calculated represent the market volatility in those years.

PLEXOS uses Monte Carlo simulations to model random events such as unplanned plant outages. As these events are random, the simulations will result in a range of prices. For example, one sample run of 2008/09 for 50% POE may result in an annual price of $36/MWh and another sample, which may have modelled unplanned outages during peak periods may result in an annual price of $39/MWh. The PLEXOS results that were used in the analysis were an average across simulations. The standard deviation of the PLEXOS results represents the variation in the PLEXOS results that occur due to the modelling of random events.

The analysis of off-peak and peak contract positions using PLEXOS modelled prices took into account the standard deviations due to the modelled NSW market volatility in PLEXOS and the price variations between PLEXOS simulations. These standard deviations were calculated using Equation C1.

2)^Pr(2)^( icesRLRStdevlationsPlexosSimuStdev + Equation C1

Where Stdev(PlexosSimulations) represents the price variations between PLEXOS simulations and Stdev(RLRPrices) represents the modelled NSW market volatility as calculated from the equation derived between the Regulated Load Rank and the natural logarithm of the NSW price. This concept is somewhat heuristic, but the intention was to characterise the uncertainty of price for a given regulated load level as the combination of two uncertainties, that of the market itself and that of the relationship between the average price and the load level. There is scope for further analysis and refinement of this approach.


APPENDIX D NET SYSTEM LOAD PROFILE

D.1 Analysis of Historical Data 2003 to 2006 EnergyAustralia provided historical data for the estimated net system load profile plus the controlled load profile together with hourly temperature data (including some data gaps) for the period from July 2003 to June 2006. Load data from January 2002 was also available but not the temperature data. In order to assess the trends in the net system load, MMA examined the pattern of this load in terms of daily energy, peak daily energy, peak demand and the key drivers of demand being:

• Load growth which was taken as a constant exponential (geometric) change on a financial year basis for the purpose of identifying the other factors

• Whether or not the day was a working week day

• Whether or not the day was week day public holiday

• An autoregressive integrated moving average (ARIMA) function of heating degree days with the threshold temperature changing over time and using an exponential weighting between daily periods

• An ARIMA function of cooling degree days with the threshold temperature changing over time and using an exponential weighting between daily periods.

Table D- 1 summarises the modelled function and the key terms which were applied to modelling the daily energy, the daily peak energy and the daily peak demand.

Figure D- 1 shows the regression fit of the daily energy for the three financial years from 2003/04 to 2005/06. The temperature and heating and cooling degree day method explains a good deal of the annual variation in daily energy demand without a specific relationship to day of the week or time of the year. The major errors occur in the changing seasons such as in September and October and April. The regression was conducted over the whole period so that we could assess the time related trends. The assessed parameters are shown in Table D- 2.

Table D- 2 shows the factors that were derived and provides an analysis. The interesting features that were assessed were as follows.

• The working week day demand and the weekend energy are maintaining a constant relativity.

• The impact on public holidays is unclear because the number of samples was small. The energy and peak demand on these days seems to be slightly decreasing relative to weekends and working days.

• We did not separate out the Christmas period from other holidays but there would be a good basis for doing so if it were important to obtain a more accurate assessment of the effect. It was not critical for this project.


Figure D- 1 Regression of Daily Energy


15000200002500030000350004000045000500005500060000

1/07

/200

4

15/0

7/20

04

29/0

7/20

04

12/0

8/20

04

26/0

8/20

04

9/09

/200

4

23/0

9/20

04

7/10

/200

4

21/1

0/20

04

4/11

/200

4

18/1

1/20

04

2/12

/200

4

16/1

2/20

04

30/1

2/20

04

13/0

1/20

05

27/0

1/20

05

10/0

2/20

05

24/0

2/20

05

10/0

3/20

05

24/0

3/20

05

7/04

/200

5

21/0

4/20

05

5/05

/200

5

19/0

5/20

05

2/06

/200

5

16/0

6/20

05

30/0

6/20

05

Day

MW

h

FitActual


15000

2000025000

30000

350004000045000

5000055000

1/07

/200

5

15/0

7/20

05

29/0

7/20

05

12/0

8/20

05

26/0

8/20

05

9/09

/200

5

23/0

9/20

05

7/10

/200

5

21/1

0/20

05

4/11

/200

5

18/1

1/20

05

2/12

/200

5

16/1

2/20

05

30/1

2/20

05

13/0

1/20

06

27/0

1/20

06

10/0

2/20

06

24/0

2/20

06

10/0

3/20

06

24/0

3/20

06

7/04

/200

6

21/0

4/20

06

5/05

/200

6

19/0

5/20

06

2/06

/200

6

16/0

6/20

06

30/0

6/20

06

Day

MW

h

FitActual


20000

25000

30000

35000

40000

45000

50000

55000

1/07

/200

3

15/0

7/20

03

29/0

7/20

03

12/0

8/20

03

26/0

8/20

03

9/09

/200

3

23/0

9/20

03

7/10

/200

3

21/1

0/20

03

4/11

/200

3

18/1

1/20

03

2/12

/200

3

16/1

2/20

03

30/1

2/20

03

13/0

1/20

04

27/0

1/20

04

10/0

2/20

04

24/0

2/20

04

9/03

/200

4

23/0

3/20

04

6/04

/200

4

20/0

4/20

04

4/05

/200

4

18/0

5/20

04

1/06

/200

4

15/0

6/20

04

29/0

6/20

04

Day

MW

h

FitActual


Table D- 1 Parameters defining pattern of regulated load (daily energy or peak)

Term Purpose Primary parameter Secondary parameter

Constant Term

Describes the daily value applicable on weekends

A constant representing the base demand at 1 July 2003

Growth modelled by the Annual Growth factor

Week day Extra energy used on working week days

A constant representing the change on working week days

A linear time dependent term which adds to the constant each year

Holiday Extra energy used on week day public holidays

A constant representing the change on holidays

A linear time dependent term which adds to the constant each year

Annual growth

Scales the estimated energy or peak according to an annual exponential growth rate

Exponential growth factor for each financial year

Heating Load

To represent demand that is related to cool temperatures for heating purposes.

Heating Threshold Temperature defines temperatures below which a heating related demand is calculated proportional to the “heating degree days”

Heating Inertia

An exponential decay of heating contribution from previous days cool weather

Cooling load To represent demand that is related to high temperatures for cooling purposes. It may represent refrigeration load or air-conditioning.

Cooling Threshold Temperature defines temperatures above which a cooling related demand is calculated proportional to the “cooling degree days”

Cooling Inertia

An exponential decay of cooling contribution from previous days hot weather


Table D- 2 Parameters for Net System Load plus Controlled Load Pattern

July 2003 Parameter

Daily Energy Weekday Peak Energy

Daily Peak Demand

Trend over Time and Comments

Constant (Standard Week end day)

28,168 MWh 22,178 MWh 1,472 MW

Working Week day

2,614 MWh N/A 144 MW

Rate of change 0 MWh/year N/A -17 MW/year 10% - 20% per annum

Week day Holiday

-946 MWh N/A -5.9 MW

Rate of change -111 MWh/year N/A -5.3 MW/year

Generally a negative trend but insufficient samples to be clear

Heating Threshold

15.59 C 15.62 C 16.14 C Trans Grid uses 18 C

Rate of change 0.50 C/year 0.50 C/year 0.50 C/year Rate was capped to 0.5C/year.

Heating Degree Day

1,310 MWh/HDD 1,045 MWh/HDD

82 MW/HDD

Rate of change 0 MWh/HDD/year

114 MWh/HDD/year

1 MW/HDD/year

11% / year for peak energy. No change for daily energy.

HDD decay ratio per day

0.701 0.609 0.727 Lag is more important for peak. Peak effect depends more on tomorrow’s weather!

Cooling Threshold

22.14 C 22.07 C 22.95 C TransGrid uses 21 C.

Rate of change -0.40 C/year -0.50 C/year -0.50 C/year Decreasing slowly. Rate of change for weekday energy was capped to -0.5C/year


July 2003 Daily Energy Weekday Daily Peak Trend over Time and Parameter Peak Energy Demand Comments

Cooling Degree Day

1,183 MWh/CDD 981 MWh/CDD

113 MW/CDD

Rate of change 254 MWh/CDD/year

337 MWh/CDD/year

24 MW/CDD/year

High levels of change was assessed

CDD decay ratio per day

0.445 0.383 0.181 Lag is more important for total daily energy

2003/04 Exponential Growth

-5.46% -8.57% -3.86% Daily peak is decreasing to a lesser extent than energy throughout the period.


-11.35% -15.61% -9.84%


-4.54% -7.96% -4.71%

• The heating threshold is similar for energy and peak demand at about 16 C in mid 2003 and the values are increasing over time by about 0.5 C per year as shown by the lower lines in Figure D- 2. This may reflect increasing affluence and the ability to buy comfort. This increase has a substantial affect on heating load because a 1 C threshold increase has the effect of increasing the Heating Degree Days for daily energy by between 25% and 30% depending on the weather. The threshold used by TransGrid for estimating heating load at the system level is 18 C according to the 2006 Annual Planning Statement. This value is about current value for 2007 for the net system load. The trend is heading past that value and will surpass it in 2008 if the current trend continues. The evident fact that the heating and cooling thresholds meet in 2010 at 19 C is addressed below. Clearly it would be unlikely that the thresholds would continue to change in this fashion indefinitely. Further work is warranted to quantify a transition to new stable threshold temperatures.


Figure D- 2 Trend in Threshold Temperatures

Trend in Threshold Temperatures

15

16

17

18

19

20

21

22

23

24

2003 2004 2005 2006 2007 2008 2009 2010

July

Degr

ess

Cel

sius

Peak Demand SummerPeak Energy SummerDaily Energy SummerPeak Demand WinterDaily Energy WinterPeak Energy Winter


• The heating degree day sensitivity of the weekday daily energy is increasing at about 11% per year which increases the low temperature sensitivity of the energy demand. It reflects increased application of electric heating which would be expected to be derived from increased use of reverse cycle air-conditioners. Figure D- 3 shows that the increasing in heat load is predominantly in the peak period with the off-peak period heating driven more by the increase in the heating threshold as seen in Figure D- 2.

• The heating lag is quite significant at 0.701 for energy and 0.727 for peak demand. This means that a series of cold days can build up a substantial heating load.

• The daily peak demand was able to be approximated more accurately by using the HDD ARIMA function for the following day. This is consistent with the TransGrid analysis in the 2006 Planning Statement which states that peak demand in winter is affected by the overnight temperature following the peak.

• The initial cooling threshold is similar for energy and peak demand at about 22 C and it is decreasing slowly over time by about 0.5 C per year as shown by the upper lines in Figure D- 2. This reflects increasing affluence and ability to buy comfort with air-conditioning. This compares favourably with the threshold used by TransGrid for estimating cooling load at the system level which is 21 C according to the 2006 Annual Planning Statement. The regression analysis for the trend in threshold temperatures also required a constraint on the rate of change to obtain credible results. Further work on this aspect of the cooling patterns is needed to be able to better estimate future trends.


Figure D- 3 Trend in Sensitivity to Heating/Cooling Degree Days

Trend in Sensitivity to Degree Days

0

500

1000

1500

2000

2500

3000

3500

4000

2003 2004 2005 2006 2007 2008 2009 2010

July

MW

h pe

r Deg

ree

Day

0

50

100

150

200

250

300

MW

per

Deg

ree

Day

Peak Cooling

Daily Cooling

Daily Heating

Peak Heating

Peak DemandCooling (RHS)

Peak DemandHeating (RHS)


• A 1 C reduction in the cooling threshold increases the amount of Cooling Degree Days by between 33% and 38% based on the 2003 to 2006 data model. The reduced threshold and the increasing sensitivity indicate that cooling load is increasing as a relative proportion of the demand.

• According to data provided to MMA by EnergyAustralia, the penetration of air-conditioner has increased from 47% in 2003 to 57% in 2005 at 5% per year. The penetration is forecast to increase to 75% by 2010, which is 3.6% additional penetration per year.

• The Cooling Degree Day sensitivity of the daily energy is increasing at about 21% per year which increases the high temperature sensitivity of the energy demand. It reflects increased penetration and application of air-conditioners. The assessed increase in penetration of air-conditioners by EnergyAustralia is about 5% per year. From a base value of 47% in 2003. This makes the annual increase 5%/47% = 10.6% of the 2003 year base value. This goes part of the way to explain the 21% rate on the 2003 year base. The remainder of the increase could be explained if new installations were 30% larger than existing installations due to the lower cost of the new equipment. We have not attempted to adjust the trend for the expected lower rate of penetration as the basis to do so is unclear.

• The peak demand sensitivity for heating is quite stable but the overall demand would be increasing with the increase in the HDD threshold. The stable sensitivity is shown in Figure D- 3.


• The cooling lag is less significant than for heating and is at 0.445 for energy and 0.181 for peak demand. This means that a series of hot days can build up the cooling energy load but it has a lesser impact on the peak demand. This would reflect higher duty on the air-conditioning plant in the morning and early evening but little change in duty cycle at the afternoon peak.

D.2 Application of the Model These relationships have been projected forward to derive a load shape for the net system load plus controlled load profile assuming that the trend of the last three years continues. One adjustment that seemed warranted was to decrease the projected peak period energy by 10% below that derived from the regression. This was necessary because the peak energy/weekday energy became too high on some days which was not realistic. It was not possible in the time available to fully resolve why this might have been occurring. The change was supported by an examination of the projected peak energy/total energy trend and the annual load factor as shown in Figure D- 4. Adjusting the peak energy gave a consistent trend projection for the peak/total energy ratio in the future and continued the historical trend with some slowing as would be expected as air-conditioning load tends to saturate.

Figure D- 4 Load Factor and Peak/Total Energy ratios – Actual and Forecast

NSL Peak Energy / Total Energy Trend

40.0%

42.0%

44.0%

46.0%

48.0%

50.0%

52.0%

2003

/04 2004

2004

/05 2005

2005

/06

2006

/07

2007

/08

2008

/09

2009

/10

Peak/Total EnergyAnnual Load Factor


The load factor outcome is more difficult to assess because the load factor has been low in the past (except 2005/06) due to the decline in total energy volume of up to 15% per annum. This has had the effect of reducing the assessed annual load factor because energy consumption falls after the annual peak has been attained. However the trend on peak demand relative to total energy restores the trend observed from 2003/04 to 2004/05. Overall this analysis indicates that the peak adjustment is reasonable to obtain an adequate representation of future load shape. Further refinement of the trend on the temperature thresholds would further enhance the accuracy with which future load factors can be assessed.

The market modelling is based upon the 2005/06 weather which was amended to represent 90%, 50% and 10% POE versions by adjusting the average temperature on the peak winter and summer day as per the advice in the 2006 TransGrid Planning Statement. The weighted temperatures were calculated and then the hourly temperatures in the critical days were scaled between the maximum and minimum temperature so that the weighted index defined by TransGrid was obtained for the peak day or days. An example of this scaling is shown in Figure D- 5 for 12 August 2005 and 18 February 2006 to obtain the required profiles.

For winter TransGrid uses the daily maximum temperature and the following minimum overnight temperature. There is a weighting of 65% for today’s temperature, 25% of the previous day and 10% from the day before that.

For summer, TransGrid uses the daily maximum temperature to 3pm, and the daily minimum temperature to 9am on that day. There is a weighting of 85% for today’s temperature, and 15% for the previous day’s temperature.

On this basis the critical temperatures are as shown in Table D- 3.

Table D- 3 Critical Weighted Daily Temperature for Sydney

90% POE 50% POE 10% POE

Sydney Winter 11.0 10.0 9.1

Sydney Summer 27.3 29.6 32.3


Figure D- 5 Example of scaling of hourly temperatures to obtain the critical profile.

Temperature Adjustments for Peak Days - 12 August

0

2

4

6

8

10

12

14

16

18

20

0:00 4:48 9:36 14:24 19:12 0:00

Time of Day

Cel

sius

Actual 200590% POE50% POE10% POE

Temperature Adjustments for Peak Days - 18 February

15

20

25

30

35

40

0:00 4:48 9:36 14:24 19:12 0:00

Time of Day

Cels

ius

10% POEActual 200650% POE90% POE

D.3 Daily Half-hourly values Once the daily total energy, peak energy and peak demand were obtained, the half-hourly values were obtained by a technique which:

• Selected the nearest historical weekday profile for the same day


• Calculated the load duration curve for that historical day between the minimum load and the maximum load

• Reshaped the load duration curve by taking 1 – (1 – Load Rank)^Power Index where Power Index varied between 1.5 and 0.5. The Power Index was selected together with a minimum load to fit the daily energy and the peak energy on weekdays using a least squares method. This method reshapes the load duration curve mostly at the top of the load curve and has less impact on the off-peak shape.

• Maintained the daily peak to match the forecast from the regression model forecast.

The method did not exactly fit the required energy profiles and some further work would be desirable to refine the reshaping of the load duration curve in peak and off-peak periods to better match the forecast. An example of the input and output load shapes for 2008/09 winter and summer is shown in Figure D- 6. The daily profiles have been shifted by three days so that they are compared for the same day of week and weather from the original 2005/06 profile. The profiles show an increase in the peakiness of the winter demand and additional cooling load on summer days in accordance with the regression trend for the energy. Whilst the shapes are not an exact fit to the daily energy profile, the errors were small, at about 1% on the day and the profiles do seem to fit with what would be expected if cooling becomes a greater portion of the summer load and heating in the winter.

The changes in shapes are consistent with increased use of air-conditioners throughout the summer day with peakier usage of heating over the winter evening peak. This combination of changes is reflected in the decreasing load factor as shown in Figure D- 4 above.


Figure D- 6 Example of reshaped load profiles

Winter

0

200

400

600

800

1000

1200

1400

1600

01-Jul 02-Jul 03-Jul 04-Jul 05-Jul 06-Jul 07-Jul 08-Jul 09-Jul 10-Jul 11-Jul0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60InOut

Summer

0

200

400

600

800

1000

1200

1400

2-Feb 3-Feb 4-Feb 5-Feb 6-Feb 7-Feb 8-Feb 9-Feb 10-Feb 11-Feb 12-Feb0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

InOut

These revised load profiles were then applied to the contract analysis model to determine the required trading strategy.


energyaustralia · 3.3 mmne customer acquisition costs 27 3.4 mmne retail costs 28 3.5 mmne retail...

Documents