The Retail Electricity Market for Households and Small Businesses in Victoria

The retail electricity market for households and small businesses in Victoria

Analysis of offers and bills

July 2017

Executive Summary

This Report examines the electricity retail market in Victoria from the perspective of the offers that are made to households and small businesses. It also examines a sample of 686 household electricity bills for electricity purchased over the period from December 2016 to April 2017.

The main objective of this Report is to establish quantitative evidence of whether the retail market is delivering outcomes that are in customers’ interests. The three main strands of analysis in this Report are retail electricity prices, the retailers’ charge for their service of retailing electricity, and the savings that customers might obtain by switching to other retail offers.

Prices

The price of electricity to customers depends on many factors but most notably their level of consumption, the pattern of their consumption (if they are on time-variant tariffs), their location and the prices in their contracts with their retailer. The analysis of retailers’ offers shows a wide range of prices even for the same customer profile. The analysis of the sample of bills also shows a wide range between highest and lowest although the median price amongst the popular retailers is similar.

The average price of electricity for the cohort of customers in the sample consuming 3.75% above and below 4,000 kWh per year, is 34.7 cents per kWh (after GST). This is 15% higher than the estimate of the representative electricity price for households in Victoria, produced by the Australian Energy Markets Commission (AEMC)1. The difference is likely to be partly explained by the AEMC’s assumption that customers are always on their retailers’ cheapest offers.

1 A like-for-like comparison of average prices charged by the Big Three retailers finds that they are 22% higher in the sample than the AEMC’s latest estimates – see page 57.

Retailers’ charges

The amount that retailers charge for their services is not separately itemised on customers’ bills. Instead it has to be deduced by subtracting the known or estimated charges for the other elements (the purchase of wholesale electricity, charge for network and metering and environmental levies) that together with the retailers’ charge make up the bill that the customer is charged. The breakdown of the bill for the representative customer in the sample is shown in Figure E1.

Figure E1: Disaggregation of average household electricity bill from sample

Representativehousehold electricity%bill%from%sample% $1,600

$1,400 $126% $18% $55% $88% $1,200 $263%

$1,000 $415% $800

$600 $423% $400

$200

$0 Retailer's%%%%%%%%%%%%%%%%%Network%%%%%%%%%%%Wholesale%%%%%%%%%%%%%%Metering%%%%%%%%%%%%% Federal% Victoria% GST charge charge charge charge environmental environmental

Looking at the estimate of the retailers’ charge across the sample, we find that for three- quarters of customers, the retailers’ charge is more than the wholesale charge, and for households in three of the five distribution areas, the retailers’ charge is higher than the network charge for most customers.

The retailers’ charge in Victoria is remarkable by comparison with the estimated retailer charge in other European countries, many of which have fully or partially deregulated retail energy markets. Figure E2 below presents this comparison where the retailers’ charge (on the Y-axis) is presented in cents per kWh.

Figure E2: Cross-country comparison of retailers’ charges for their retail service, to residential customers (cents per kWh)

12.0%

10.0%

8.0%

6.0%

4.0%

2.0% Retailers'%charge%%(Australian%cents%per%kWh) ! Italy Spain France Poland Ireland Austria Greece Estonia Finland Holland Victoria Norway Sweden Slovakia Belgium Slovenia Portugal Hungary Germany Denmark Lithuania EU%Average Queensland Luxumbourg Great%Britain South%Australia New%South%Wales

While the range of the retailers’ charge in the sample bills varies considerably, the retailers with the greatest number of bills in the sample (corresponding to their market share in the population) tended to have the highest retailer charge as a proportion of their customers’ bills, as shown in Figure E3 below:

Figure E3. Distribution of retailer charges as a percentage of total bill (after GST), by retailer

For most households in the sample, the retailers’ charge for its services is the biggest single component of their electricity bill. About three out of four households in the sample are paying more for electricity to be sold to them by their retailers than they are paying for it to be produced by generators.

Comparing the retailers’ charge with the charge for network services, the analysis finds that in three out of the five distribution regions in Victoria, most households are being charged more for electricity to be sold to them than they are paying for it to be transported over the transmission and distribution network.

The estimate of the retailers’ charge in the sample of bills is affected by assumptions of the wholesale price. The conclusion that the retailers’ charge is more than the wholesale charge is robust to far higher estimates of the wholesale price than the one used in this report.

Saving by switching

A well established narrative in the discourse on retail energy markets in Australia is that customers can save significantly by switching supplier. This report tested this by analysing the savings that customers in the sample would receive if they switched to the least expensive offer. The analysis found that customers in the sample could, on average, reduce their bills by $294 per year, or around 21% by switching to the least expensive offer. Segmentation of the savings into three clusters is shown in Figure E4.

Figure E4: Switching savings in three clusters

• The “Low saving” cluster accounts for 204 out of the 686 bills. The median saving for customers in this cluster is $84 per year. • The “Moderate saving” cluster has 280 bills and a median saving of $223 per year. • The “High saving” cluster has 155 bills with median saving of $501 per year.

In addition to bills in these three clusters, there were 33 bills that were cheaper than any offer in the market and there were 8 bills with savings of more than $1,000.

The analysis of switching savings also shows that customers served by the retailers with the greatest number of customers in the sample will tend to save more than customers of other less popular retailers.

This analysis assumes that customers are able to identify the least expensive offer for themselves taking account of their consumption profile, whether or not they have

6 controlled load or solar, their location and their tariff type. It also assumes that customers are able to access and evaluate all the relevant competing offers and that the cheapest offers for them remain available for at least a year (or that customers switch to comparably cheap offers if the chosen offer changes).

These are onerous assumptions and so the analysis tested the savings that customers might obtain if they switched to the second, third, fourth and up to the tenth cheapest offer. This found that the savings reduced considerably so that if the customer switched to the fifth cheapest offer, the median saving in the sample would be less than half what it would be if they switched to the cheapest offer. This suggests that the narrative that customers can reduce their bill by switching should be tempered by the evidence that the extent of savings depends greatly on customers’ ability to identify and switch to the cheapest offer – and continue to switch again if the cheapest offer changes.

Finally the analysis of switching saving considered the extent to which customers might reduce their bills by switching to the least expensive offer from their own retailer. This also showed a wide range of savings for bills in the sample. However, the median saving that customers might achieve by switching to the lowest offer from their own retailer is considerably lower than the saving that they might achieve by switching to the lowest offer in the market.

Table of Contents

1 Introduction 13

Part A

2 Description of the retail market 15 2.1 Customer numbers 15 2.2 Retailers and retail offers 16 2.3 Fixed versus variable charges in retail offers 22 2.4 Discounts 24 2.5 Incentives 27 2.6 Network versus retail charges 27 2.7 The comparison challenge 30

3 Bill disaggregation based on retail offers 34 3.1 Method 34 3.1.1 Estimating the bill 34 3.1.2 Estimating the wholesale charge 37 3.1.3 Estimating network charge 40 3.1.4 Estimating environmental charges 40 3.1.5 Metering charges 41 3.1.6 GST 41 3.2 Results 41

4 Inter-state and international comparison of prices and retailer charges in residential offers 43 4.1 Prices 44 4.2 Retailer charges 44

Part B

5 Sample data description 46

6 Retailers’ charges in the sample bills 58

7 Switching savings 71

7.1 Savings by switching to the least expensive offer in the market 72 7.2 Cluster analysis of saving 76 7.3 Impact of ability to identify cheapest offer 77 7.4 Savings by switching to the least expensive offer from the existing retailer 78

Appendix A: Data and analytical tools 82

Appendix B: Load profile assumptions 83

Table of Figures

Figure 1. Box plot of annual charge ($)by retailer assuming 4 MWh per year ...... 19 Figure 2. Box plot of annual charge ($)by retailer in Powercor area assuming 4 MWh per year ...... 19 Figure 3. Analysis of residential time of use tariffs (market offers) ...... 20 Figure 4. Offer duration: number of days before 14 May 2017 when offers were introduced to the market (all market offers, residential and small business) ...... 21 Figure 5. Scatter plot of fixed versus total charge in residential market offers (assuming 4 MWh annum household) ...... 23 Figure 6. Scatter plot of fixed and variable charges in residential market offers ... 23 Figure 7. Histogram showing frequency of annual charges (before-GST) with and without conditional discounts (assuming a 4 MWh per year household) ...... 26 Figure 8. Histogram showing frequency of annual charges (before-GST) with and without conditional discounts (assuming a 10 MWh per year small business) ...... 26 Figure 9. Total retail charge minus total network charge ($ per year) 4 MWh per annum household ...... 28 Figure 10. Variable retail charge and variable network charge ($ per year) 4 MWh per annum household ...... 28 Figure 11. Fixed retail charge and fixed network charge ...... 29 Figure 12. Annual fixed retail and fixed network charge less metering charge ...... 29 Figure 13. Offer duration, all market offers, residential customers at 30 May 2017 ...... 38 Figure 14. Residential bill disaggregation based market offers, 4 MWh p.a...... 42 Figure 15. Small business bill disaggregation, 10 MWh p.a...... 42 Figure 16. Comparison of residential electricity prices (before and after tax) (Australian cents per kWh) (May 2017 prices in Australia, 2015 prices in European countries) ...... 44 Figure 17. Inter-state comparison of retailer charges to residential customers ...... 45 Figure 18. Comparison of retailer charges to residential customers (cents per kWh) ...... 45 Figure 19. Distribution by tariff type and network service provider ...... 47 Figure 20. Distribution of tariff type by retailer ...... 47

Figure 21. Density distribution of daily consumption by cluster and in aggregate ...... 48 Figure 22. Fixed charges versus total bill ($ per year) ...... 50 Figure 23. Annual fixed retail, fixed network and metering charges ...... 50 Figure 24. Distribution of discounts by network service provider ...... 51 Figure 25. Description of discounts ...... 52 Figure 26. Switching saving: impact of discount rate ...... 53 Figure 27. Description of retailers’ solar feed-in rates ...... 54 Figure 28. Distribution of the estimated annual bills ($ per customer per year) in each distribution zone ...... 56 Figure 29. Retailers' charge ($ per year) versus wholesale price ($/MWh) ...... 60 Figure 30. Residential bill disaggregation based on average bill from sample (4 MWh) ...... 63 Figure 31. Histogram of the difference between network and retailers' charges ..... 64 Figure 32. Histogram of the difference between wholesale and retailers' charges ...... 64 Figure 33. Histogram of retailers' charge by network service provider area ...... 65 Figure 34. Distribution of retailer charges as a percentage of total bill (before GST), by retailer ...... 67 Figure 35. Cross-country comparison of retailers' charge (cents per kWh) using sample of bills ...... 68 Figure 36. Ratio of retailers' charge to wholesale charge ...... 70 Figure 37. Distribution of annual saving by selecting the cheapest offer, by network service provider area ...... 73 Figure 38. Distribution of saving (as a percentage of total bill) by selecting the cheapest offer, by network service provider area ...... 73 Figure 39. Impact of controlled load on saving (as a percentage of the total bill) by selecting the cheapest offer ...... 74 Figure 40. Impact of solar on saving (as a percentage of the total bill) by selecting the cheapest offer ...... 75 Figure 41. Distribution of savings by cluster ...... 76 Figure 42. Switching savings in three clusters ...... 77 Figure 43. Distribution of saving ($ per customer per year) by switching to cheapest market offer versus lowest existing retailer offer ...... 79 Figure 44. Median saving ($ per customer per year) by switching to the lowest market offer versus lowest offer from customers’ existing retailers ...... 80

Figure 45. Relationship between switching saving and retailers’ market share in actual bill sample ...... 81

Table of Tables

Table 1. Actual number of customers by tariff type - residential ...... 16 Table 2. Actual number of customers by tariff type – small business ...... 16 Table 3. Number of offers (market and standing) to residential and small business by distribution area ...... 17 Table 4. Number of offers (market and standing) residential customers ...... 18 Table 5. Variable charges in flat and block offers to residential customers (market offers) ...... 20 Table 6. Variable charges in flat and block offers to small business customers (market offers) ...... 20 Table 7. Summary statistics on daily charges in residential retail offers ...... 22 Table 8. Summary statistics on daily charges in small business retail offers ...... 22 Table 9. Summary statistics on discounts in market offers to residential customers ...... 24 Table 10. Summary statistics on discounts in market offers to small business customers ...... 25 Table 11. Retailer market share in sample compared to population ...... 49 Table 12. Summary information of discounts by type and incidence ...... 52 Table 13. Summary statistics of estimated annual bills ($ per customer per year) ...... 55 Table 14. Summary statistics of estimated annual prices (cents per kWh) ...... 57 Table 15. Analysis of average and median bills in the sample ...... 61 Table 16. Summary statistics on retailer charges for their services ($ per customer per year) ...... 66 Table 17. Summary statistics of estimated saving from switching to the lowest market offers ($ per customer per year) ...... 72 Table 18. Relationship between switching saving and selection of alternative offers ...... 78

1 Introduction

This Report provides quantitative information on the electricity retail market in Victoria using data from publicly available offers and from a sample of customers’ bills. The Report covers the retail market applicable to residential and small business customers.

This Report has two main parts:

• The first part (in Sections 2, 3 and 4) analyses the offers that retailers make to new customers, in order to estimate bills, prices and retailers’ charges for their services and compare these nationally and internationally. • The second part (in Sections 5, 6, and 7) analyses a sample of 686 residential electricity customer bills. The analysis of these bills provides information based on what customers are actually paying and answers other questions, such as how much retailers are charging for their services and how much customers could save by switching to the cheapest retailers.

To delineate these two separate strands of analysis, Sections 2, 3 and 4 form Part A, and Sections 5 to 7 form Part B. The Appendices describes the tools used to undertake the analysis.

The graphical representations used in this report are common in data science and statistics but are not always widely used in energy economics and policy studies in Australia. A summary of the graphical depictions used in this Report are as follows:

• Histogram: this depicts the distribution of the data by displaying the number of data items with values corresponding to the count of those data items (on the x- axis) corresponding to the value depicted on the Y-axis. • Box plots: this depicts the range of the chosen data item by drawing the box that shows the values between the 25th and 75th percentile. The median of the range is shown as a bar in the box and “whiskers” in the box show the range of the data 1.5 times the difference in the 25th and 75th percentile.

• Dot plots: this shows the median value of the range as the dot and the bars either side of dot as the limits of the range of the data. • Scatter plot: this plots the relationship between two variables. Where applicable, we have included lines of best fit of the data based on linear regressions. Where applicable we also show the “R-squared” error which measures how well the line of best fit, fits the data (an R-squared error of 1 is a perfect fit).

2 Description of the retail market

This section describes the Victorian retail electricity market. It describes, in order, customer numbers, retailers and retail offers, fixed versus variable charges in retail offers, discounts, and retail versus network charges.

2.1 Customer numbers

There are around 2.4 million residential electricity accounts and 340,000 small business electricity accounts. A wide variety of tariff structures are used and these can be summarised under five headings:

• Block tariffs: These are tariffs which charge different amounts for blocks of consumption metered daily, monthly or three monthly. There may be between 2 and 4 blocks. Some retailers have seasonal block charges but this is rare. A daily charge is also levied. • Flat tariffs: This tariff has a single rate for consumption and a daily charge. • Time of Use: This is a tariff that has different charges in two periods. For residential customers the peak period is from 7am to 11pm Monday to Friday. For business customers peak is from 8am to 10pm seven days a week. Some retailers use different time definitions but this is rare. Some retailers charge for the peak period in varying blocks, but this is rare. A daily charge is also levied. • Flexible tariffs: This is a tariff with charges in peak, shoulder and off-peak periods. Some retailers charge for the peak period in varying blocks, but this is rare. Flexible tariffs can have seasonally varying charges but this is rate. A daily charge is also levied.

Some customers also have dedicated circuits for the purpose, mainly, of water heating. The dedicated charge is a single cents per kWh rate. In addition to these tariffs, tariffs that charge for household peak demand have been introduced but there are almost no customers on these tariffs. Table 2show the number of residential and small business customers classified by tariff type. The main observation in relation to residential customer is that around 2 million of the 2.3 million accounts are flat or block tariffs and of this 2 million around 15% also have dedicated circuit tariffs.

Table 1. Actual number of customers by tariff type - residential

Block or flat Block or without flat with Two-part Dedicated dedicated dedicated time-of- circuit circuit circuit use Flexible TOTAL Ausnet 99,322 391,368 99,322 62,873 - 652,885 Jemena - 275,000 - 16,927 1,100 293,028 United 13,300 490,566 13,300 24,702 - 541,868 Powercor 175,123 351,312 175,123 161,899 76 863,534 Citipower 27,052 207,504 27,052 41,342 23 302,973 TOTAL 314,797 1,715,751 314,797 307,744 1,199 2,339,490

The main observation in relation to small business customers is that a little more than half have time-varying offers. For both residential and small business customers, flexible (three part time of use) tariffs have a small number of customers.

Table 2. Actual number of customers by tariff type – small business

Single Time-of- rate use Flexible TOTAL Ausnet 27,113 81,200 3,226 111,539 Jemena 15,361 11,237 - 26,599 United 37,017 18,236 - 55,253 Powercor 40,081 48,835 - 88,916 Citipower 25,007 28,772 - 53,779 TOTAL 144,578 188,280 3,226 336,085

2.2 Retailers and retail offers

Twenty-four retailers offer to sell grid-supplied electricity to households and/or small business electricity consumers in Victoria. Almost all of these retailers make “market” offers and “standing” offers. Retailers have discretion in setting the prices in these offers, but are restricted from changing their standing offers more than twice per year. Retailers are required to post the terms of their offers in electricity price fact sheets on their websites. In total on 11 May 2017 there were 2,841 offers to residential and small business customers in the five regions of Victoria on the Victorian Government’s Switchon price comparison website as summarised in Table 3 below.

Table 3. Number of offers (market and standing) to residential and small business by distribution area

Retailer / network service united provider ausnet citipower jemena powercor energy agl 23 18 24 30 26 alinta 5 6 6 5 6 bluenrg 9 9 9 7 9 click 33 32 33 21 33 commander 25 19 20 25 24 covau 14 15 14 8 14 diamond 9 11 10 9 9 dodo 11 11 12 15 14 ea 18 18 18 18 18 erm 9 9 10 8 7 firstenergy 12 12 12 12 11 globird 14 27 16 14 13 lumo 37 37 37 37 37 momentum 37 37 30 39 41 next 12 14 13 12 11 origin 40 40 40 40 40 pacific 12 18 14 23 12 peopleenergy 6 6 6 7 6 powerdirect 6 9 7 7 7 powershop 11 12 11 8 11 qenergy 27 36 19 13 20 redenergy 7 10 10 14 10 simply 44 41 40 46 47 sumo 21 21 21 21 21 TOTAL 442 468 432 439 447

The choice of offers available to customers is often limited by their type of tariff, with many retailers not offering to change customers to a different type structure. Table 4 shows the number of market and standing offers by retailer, classified by tariff type to residential customers

Table 4. Number of offers (market and standing) residential customers

Retailer Block and Time of use Flexible flat agl 49 41 0 alinta 18 10 0 bluenrg 9 4 0 click 65 30 3 commander 25 53 0 covau 18 8 8 diamond 15 9 8 dodo 23 14 10 ea 40 20 15 firstenergy 19 10 0 globird 31 16 13 lumo 70 35 20 momentum 39 35 19 next 10 5 0 origin 80 40 80 pacific 23 24 0 peopleenergy 10 6 0 powerdirect 21 10 0 powershop 20 9 9 qenergy 21 28 6 redenergy 20 11 5 simply 64 58 10 sumo 30 15 12 sumo 45 45 12 TOTAL 765 536 230

To show the range of the offers made by each retailer and in comparison to other retailers, Figure 1 shows box plot of the annual charge in market offers that each retailer make to residential customers. The calculation in this chart assumes all conditional discounts are received. Part of the range show in these box plots is explained by differences in the various network service providers, reflecting different network service provider chargers. To adjust for this, Figure 2 shows box plots for offers in just one network service provider area (Powercor). Relative to Figure 1 it shows a narrower, but still wide range of offers from most retailers and also amongst retailers.

Figure 1. Box plot of annual charge ($)by retailer assuming 4 MWh per year

Figure 2. Box plot of annual charge ($)by retailer in Powercor area assuming 4 MWh per year

Table 5 presents summary statistics on the variable charges in flat and block tariffs for residential customers. It shows that block tariffs typically have higher rates than the variable rate in flat offers and also that the median rate in block offers barely varies between blocks. Table 6 for small business customers shows that median and average flat rates are about 10% higher than for residential customers, while block rates are about the same.

Table 5. Variable charges in flat and block offers to residential customers (market offers)

Block tariffs

Flat rate Block 1 Block 2 Balance (cents per (cents per (cents per rate (cents kWh) kWh) kWh) per kWh) Highest 35.4 41.7 33.7 38.5 Lowest 14.1 10.9 22.7 11.2 Median 25.8 27.7 28.6 28.5 Average 25.2 27.1 28.8 26.9 Table 6. Variable charges in flat and block offers to small business customers (market offers)

Block tariffs

Flat rate Block 1 Block 2 Balance (cents per (cents per (cents per rate (cents kWh) kWh) kWh) per kWh) Highest 48.8 40.9 37.7 45.3 Lowest 15.3 12.6 22.7 12.4 Median 27.5 26.6 27.8 26.6 Average 27.6 26.8 29.0 27.0

Figure 3 analyses the residential time of use offers. It shows that the off-peak rate ranges between a low of just under 10 cents per kWh to a high of more than 27 cents per kWh. It is also notable that the difference between the peak and off-peak rates varies over a very wide range.

Figure 3. Analysis of residential time of use tariffs (market offers)

10 peak rate (cents per kWh)

- 5

Off 0 0 5 10 15 20 25 30 Peak minus off-peak rate (cents per kWh)

In addition to the generally available retail offers that are posted on retailers’ website, some retailers make bespoke offers that are not generally available and so the terms of these offers are not included on their websites. An example of such bespoke offer is Origin Energy’s “Predictable Power” plan which offers fixed monthly charges for 12 months.

Retailers update their market offers frequently. Many of them change their market offers around a few dates such as the start of the year but again several other times a year. Figure 4 shows that around 25% the 1,476 market offers to households and small businesses on 14 May 2017, were introduced around the start of the year, with the remaining 75 % introduced mostly more recently. It should be noted that Figure 4 is obtained by analysing market offers based on their offer dates. We are aware of at least one instance where rates have been changed but the date of the revised offer stayed the same as the original offer. The data shown in Figure 4 might therefore be considered to be a conservative estimate of the spread of offer duration.

Figure 4. Offer duration: number of days before 14 May 2017 when offers were introduced to the market (all market offers, residential and small business)

600

500

400

300

200 Number of offers 100

0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500

No of days before 14 May 2017

2.3 Fixed versus variable charges in retail offers

Fixed charges (the daily charge) in residential electricity offers in Victoria are typically higher than elsewhere in Australia and higher than we have observed in any other electricity market. Summary statistics of the fixed charge on residential retail offers2 in Victoria is shown in Table 7 below.

Table 7. Summary statistics on daily charges in residential retail offers

Daily charge (cents per day) Highest 260 Lowest 72 Median 114 Average 114

Table 9 for small business customers shows that median and average daily rates, by comparison with Table 7 are about 20% higher than for residential customers.

Table 8. Summary statistics on daily charges in small business retail offers

Daily charge (cents per day) Highest 365 Lowest 53 Median 138 Average 146

The proportion of the customers’ bill that is accounted for by fixed charges will depend on many factors, but particularly their annual consumption and the discounts in their offers. The lower the consumption the greater the proportion of the bill that is explained by fixed charges. In addition, as described later in more detail, the predominant discount in retail offers is to discount usage charges rather than the total bill. This enlarges fixed charges relative to consumption charges. Figure 5 shows the range of fixed charges (on the y-axis) and total charge assuming that all conditional discounts in offers that have them, are met.

2 Excluding the handful of “Climate Saver” offers in the Powercor area.

Figure 5. Scatter plot of fixed versus total charge in residential market offers (assuming 4 MWh annum household)

$700

$600

$500

$400 Fixed charge $300

$200

$100 $500 $700 $900 $1,100 $1,300 $1,500 $1,700 $1,900 Total charge

Figure 6 narrows the focus to examine the scatter plot of the variable charge in flat offers (on the Y-axis) relative to the fixed charge on those same offers. It shows, interestingly that that there is no consistent relationship between fixed and variable offers: the flat (variable) rate varies over much the same range irrespective of the level of the fixed charge.

Figure 6. Scatter plot of fixed and variable charges in residential market offers

40 35 30 25 20 15 10

Flat rate (cents per kWh) 5 0 65 85 105 125 145 165 Daily charge (cents per day)

2.4 Discounts

Table 9 presents summary statistics on discounts in market offers to residential customers. It shows that of the 1,270 residential market offers in May 2017 there were 247 that did not have any discount. 738 offers had a discount on usage charges, of which almost all were conditional discounts. Similarly of the 285 offers that had discounts on the total bill, all of these were conditional (based on the way that the retailers depict their offers in Switchon)3.

Table 9. Summary statistics on discounts in market offers to residential customers

Number Median Average Highest Lowest of offers Discount on Total 285 10% 17% 34% 1% (conditional and unconditional) Of which unconditional 0 0 0 0 0

Of which conditional 285 10% 17% 34% 1% Discount on Usage 738 22% 22% 40% 0% (conditional and unconditional) Of which unconditional 75 18% 16% 22% 0% Of which conditional 663 26% 22% 40% 0% No discount 247 TOTAL 1270

Table 10 provides summary statistics on discounts in market offers to small business customers. It shows generally similar outcomes to those for residential customers although discounts are much less prevalent in business offers although those do offer discounts typically offer discounts that are more commonly unconditional, at least relative to what they offer to residential customers.

3Specifically, Powershop includes an unconditional discount in its Standard Saver offers’ fact sheets but it factors this in to its price and only shows a conditional discount in its depiction of these offers in Switchon.

Table 10. Summary statistics on discounts in market offers to small business customers

Number Median Average Highest Lowest of offers Discount on Total (conditional 132 0% 0% 0% 0% and unconditional) Of which unconditional 0 0 0 0 0

Of which conditional 132 0% 0% 0% 0% Discount on Usage (conditional 397 10% 22% 42% 0% and unconditional)

Of which unconditional 162 28% 26% 42% 0% Of which conditional 235 20% 19% 40% 3% No discount 219 TOTAL 748

Discounts are clearly a significant part of most Victorian retail electricity offers. However, notwithstanding this, it is notable that the least expensive offers are not actually those with the highest discounts. For example, ranking the 10 cheapest offers for a 4 MWh per year residential customer on a flat or multiflat tariff in Citipower’s area. The two least expensive offers had no discount, the fourth least expensive had just a 1% discount. The remaining six however had an average discount of 36%. The same pattern is seen in other distribution areas for both residential and small business customers.

The conditions in conditional discounts relate most frequently to on-time bill payment and the acceptance of direct debt, or on-line only accounts or payment in advance. A wide diversity of approaches are applied by retailers in the specification of their discounts. A combination of discounts are sometimes offered. Some of the most recent offers link discount rates on consumption to the solar feed-in rate this is offered (a lower discount on consumption implies a higher feed-in rate and vice versa).

Discounts are sometimes fixed for the duration of what is often referred to as the “benefit period” at the end of which they may be varied or withdrawn. Benefit periods are typically one year, but in some offers extend to two years.

As might be expected, with such large conditional discounts, the annual bill in many residential offers will be significantly affected by whether the conditions in the discount are satisfied. This is shown in Figure 7 which shows the annual charges on the offers

25 with conditional discounts. It shows that if the conditions are not met, annual charges are typically in the range from $1,300 to $1,500 per year. If the conditions are met, the annual charge is most frequently in the range from $1,100 to $1,200 per year

Figure 7. Histogram showing frequency of annual charges (before-GST) with and without conditional discounts (assuming a 4 MWh per year household)

450 400 350 300 250 200 Freqency 150 100 50 0

all discounts received only unconditional discounts

Figure 8 replicates the analysis in Figure 7 for small business customers. Unsurprisingly the same trend for small business customers can be seen as for residential customers.

Figure 8. Histogram showing frequency of annual charges (before-GST) with and without conditional discounts (assuming a 10 MWh per year small business)

400

350

300

250

200

150 Frequency 100

all discounts received only unconditional discounts

2.5 Incentives

In addition to discounts, some retailers offer various incentives mainly to entice new customers. These incentives tend to be fleeting and only some retailers include relevant details in their fact sheets. Examples of incentives in the market at the time of writing this report include:

• A $50 reduction in your first bill if you call the retailer and they don’t attend to your call in two minutes. • Free electricity for the 12th month after joining the retailer; • A $50 reduction in the first bill if the customer joins the retailer directly rather than via a switching site. • An allocation of airmiles if the customer joins the retailer and stays with them for a defined period of time. • One cinema ticket per month for each month that the customer stays with the retailer, for the first 10 months.

These fleeting incentives are not priced in the assessment of retail prices or margins in this Report.

2.6 Network versus retail charges

As discussed in the next section, the charge for retailing electricity (the “retailers’ charge”) and the charge for network services are typically the two largest components of the residential electricity bill. A question frequently asked is how network and retail charges relate to each other. This subsection explores the data on this. Three figures are presented showing in order total retail versus total network, and then the variable retail versus variable network and finally the fixed retail versus fixed network charges. The dataset used here is the average of the median offers (assuming all discounts are achieved) for AGL, Origin Energy and Energy Australia which together supply around 70% of all Victorian households. Figure 9 shows that for the average of the median offers, these retailers adjust their retail offers to reflect the network charges so that the difference between the total retail charge and total network charge is approximately the same across all five distribution areas.

Figure 9. Total retail charge minus total network charge ($ per year) 4 MWh per annum household

$1,400 Retail Network Difference

$1,200

$1,000

$800

$600 Retail minus network $400

$200

$0 Ausnet Citipower Jemena Powercor United

Figure 10 shows, in addition that retailers set their variable charges so that their net income after subtracting variable network charges is approximately constant, and about half the level of the total variable retail charge.

Figure 10. Variable retail charge and variable network charge ($ per year) 4 MWh per annum household

$1,000 Retail Network Difference

$900

$800

$700

$600

$500

$400

$300

$200 Retail minus network (variable charge) $100

$0 Ausnet Citipower Jemena Powercor United Figure 11 shows that the difference between the income retailers obtain from fixed charges on their median offers is approximately constant across distribution region. However comparing the relative size of the orange and grey bars in Figure 10 and 11, it is clear that retailers set their fixed charges in such a way that their mark up on the network service providers’ fixed charges is far higher than their mark up on the network service providers’ variable charges.

Figure 11. Fixed retail charge and fixed network charge

$500 Retail Network Difference

$450

$400

$350

$300

$250

$200

$150

$100 Retail minus network (fixed charge)

$50

$0 Ausnet Citipower Jemena Powercor United

The difference in Figure 11 might be explained as reflecting the dominant pricing policy that is chosen. It might also be argued to reflect, in part, retailers’ approach to the recovery of consumption-invariant smart meter charges from network service providers (this charge is described in the next section). Figure 12 shows the difference between the fixed retail charge and the fixed network charge after deducting the metering charge for a single phase single element meter (the predominant residential meter). In this case the gap between this difference and the retail charge is much bigger than shown in Figure 11.

Figure 12. Annual fixed retail and fixed network charge less metering charge

Retail Network Metering# Retail#E network#E metering $500#

$450#

$400#

$350#

$300#

$250#

$200# Fixed#charge

$150#

$100#

$50#

$0# Ausnet Citipower Jemena Powercor United

It is necessary to caution against drawing a conclusion from this is that the meter charges explain why fixed retailer charges are higher in Victoria than in other states. In any market there is no “correct” way to recover the charges that retailers incur, in the structure of the charges that they set in their offers to their customers. Retailers might

29 suggest that they set their retail fixed charges to recover the cents per day metering charges that they incur. But the evidence does not support this claim: the cents per day fixed charges have not declined as the metering charges have declined. To the contrary in almost all cases they have increased. For example in 2017, the weighted average metering charge to households declined by $25.40 compared to the previous year. Yet the median fixed charge in all retail offers to households in the market on 31 December 2016 was $401.30 per year, but for all the offers in the market on 12 May 2017 it had risen to $412.45. Only a small part of this difference is explained by slightly higher fixed changes in the network charges of distributors.

It should also be recognised that some retailers offer very different tariffs from their peers although in Victoria these offers are not generally available. For example both Sumo and Origin Energy offers that might be described as “all you can eat” – a fixed payment for a period irrespective of consumption in that period. In New South Wales, Pooled Energy and Mojo both offer tariffs that have annual or monthly subscription charges in lieu of charges that would otherwise be variable with consumption.

2.7 The comparison challenge

Accurate comparison of competing offers is, even in ostensibly simple cases, in fact incredibly difficult. For customers to evaluate the market they need historic information on their consumption volumes and, if on time-variant offers, the proportion of their consumption at different times of the day. They then need to consider how this might change in assessing their future requirements. If they have solar they need to be able to reliably estimate their likely annual exports to the grid. Then they need to search the market for competing offers and evaluate them. The evaluation needs to consider:

• The rates – variable and fixed. This is non-trivial in almost all cases. Block rate tariffs – common in some distribution areas - need to take account of the volume of consumption and whether the blocks are defined daily, monthly or quarterly. • Discounts – conditional, unconditional on total bill or usage. • The value of up-front incentives and progressive rewards. • If applicable, the volume and price of Green Power (only some retailers offer it and they offer varying proportions and different combinations of fixed or consumption-variant charges.

• Solar feed-in rates (which often depend on whether customers have access to premium rates) and depend on system size in some cases, system volume in others and in some cases vary over time (higher initial rates that decline after a period) • The impact of exit fees, which in some cases change depending on the duration of the customers’ tenure with the retailer. • Whether offers are fixed or not: some offer truly fixed prices for a period. Others offer fixed discounts for a period. • Whether offers in the market are actually available to them. Some retailers post apparently attractive offers but in fact on further inquiry it becomes clear that they are only available to customers with certain Network Meter Identifiers in the relevant distribution areas. Some retailers offer tariffs without fixed charges, but again on closer inspection they are not available to new entrants (and the retailers don’t say this in their fact sheets). Some retailers will accept new customers that wish to change their tariff type (time of use versus flat or block). Others, and this seems to be the default, will not.

Victoria has a range of complex offers. This reflects many factors not least a legacy from many decades ago when electricity was sold by both the state electricity commission and numerous local government authorities and the various organisations adopted different pricing practises. While many residential customers are currently on relatively straight-forward two part (fixed and variable) rates, many much more complex variants exist.

For example some retailers make offers that have seasonally differentiated peak, shoulder and off-peak charges and the the peak charges vary depending on the level of consumption in the peak period. In Powercor’s area, some customers are offered “climate saver” seasonal charges for air-conditioners connected to separate circuits. Many retailers include rates for this offer – as a form of seasonal controlled load – but on closer inspection customers scanning the market will find that they are not (in some cases) available to new customers.

A further complication is that retailers have started to offer customers charges that include rates for peak demand. Retailers measure this in different ways and customers are likely to find it difficult to predict their peak demands over the measured period.

Analysis of historic half-hourly consumption data require sophisticated analytical skill, and may nonetheless be a poor predictor of future demand charges. 4

Accurate comparison of competing offers requires that the analysis takes account of the many ways that offers differ from one another (tariff type, existence of controlled load, discount type (conditional or unconditional and if conditional, type of condition), discount incidence (total bill or energy bill), upfront bill payment, Green Power rates, solar feed-in rates, duration of benefit periods, exit fees, other fees and charges. Even apparently innocuous differences can have a significant impact on bills and, quite understandably, none of the retailers clearly spell out all the relevant terms of their offers in bill-boards, websites, TV or radio advertisements.

Finding all the relevant information to ensure a proper assessment is not easy. Retailers are obliged to post the terms of their offers on their websites in “energy price fact sheets” but in most cases these can be difficult to locate and in some cases require user interaction (e.g. post code, whether or not the user has controlled load, or solar, whether the customer wants gas and electricity or just electricity offers and so on) to order to view or download the relevant sheet. This impedes pair-wise comparison. The fact sheets are often incomplete – for example many don’t mention solar feed-in rates or Green Power rates and often also incentives - requiring users to search for the additional information on retailers’ websites. Some parts of this problem also exist in respect of the data provided by some retailers to the Victorian Government’s price comparison website. This impedes accurate comparison of offers.

In addition, unlike in other states in the NEM, retailers in Victoria are not required to follow a standard format in their energy price fact sheets. Having worked with these

4 Understandably, the Victorian Government’s price comparison website excludes these demand charges in its calculation of the bill for those offers that include such charges. However this does render an inaccurate estimate of the charges on those offers, which customers would only find out about on further interaction with the retailers offering those tariffs.

32 fact sheets extensively, obvious and not so obvious errors are apparent. In some cases careful review and correction is needed.

It might be suggested that many customers might not be concerned to find the absolute best offer in the market, but instead settle on a better offer. This may be the case but obscure and difficult details can affect bills significantly.

3 Bill disaggregation based on retail offers

This section breaks the residential and small business offers down into their component parts to derive the retailers’ charge for their service of selling electricity. This charge is some times referred to as the “gross margin”. The first subsection describes the calculation and the second presents the results.

3.1 Method

The retailers’ charge for its services is not separately disclosed on customer’s bills. Instead it has to be calculated by estimating the electricity bill and then deducting the cost of the other components that make up the bill including the wholesale, network, environmental and metering charges. Some of these charges (environmental, network and metering charges) are known or can be accurately estimated. Others, specifically the wholesale charge, might plausibly lie in a reasonably wide range.

3.1.1 Estimating the bill

The calculations here are for a representative household. The first task is to define the volume of electricity that that household purchases, and then the price that the household pays for that volume of electricity.

Dealing first with volume, we have assumed 4,000 kWh per year. This is rounded down from the 4,026 kWh per year estimate that the Australian Energy Markets Commission used for its Price Trends report5, based on the consumption estimate for a two person household that is also supplied with gas. This consumption is similar (slightly higher) than the median household consumption on Ausnet Service’s network. The Ausnet data – the only publicly available primary dataset that we know of - provides the distribution of household customer numbers by consumption band and provides confidence that the 4,000 kWh per year estimate is a reasonable estimate to use for a representative household.

5 See for example “Retail electricity price trends report, 2016” available from the AEMC’s website.

For small business customers, some will consume less than households and others very much more. For the representative small business customer 10,000 kWh per year is assumed.

Moving to the price, a weighted median is estimated. The weights are based on the latest data on retailers’ customer numbers on Standing and Market Offers (these customer number data from the Essential Services Commission) and also on the number of residential customers in each of Victoria’s five distribution regions (from the network service providers’ Regulatory Information Notices). The weighting also includes assumptions on the proportion of customers receiving conditional discounts and those not, as described below.

The challenge in developing a plausible estimate of the bill paid by a representative customer in Victoria is to account for the number of customers on Standing Offers (whose prices are easily known) and then to arrive at a plausible weighting of the estimated bills by making reasonable assumptions on the proportion of the customers that are on market offers that paying the fully discounted prices and the proportion that receive guaranteed unconditional discounts, if any at all.

The estimate of the price on Standing Offers uses the median offer of each retailer. The estimate the price in Market Offers uses a weighted average of the median price assuming that half of all customers on Market Offers receive all conditional and unconditional discounts, and that the other half are on offers that receive only unconditional discounts6.

This approach merits further explanation. As discussed in the sub-section on discounts, most retailers in Victoria offer customers conditional introductory discounts for defined periods, which are often called “benefit periods”. These benefit periods are typically a year, sometimes two, counted from the time that they have taken up the offer. The

6 This approach is equivalent to assume that in the market half of customers receive discounts equivalent to the offers made to new customers and the other half have either lost their discounts as their introductory discounts have lapsed or they or have failed to meet the conditions of the conditional discounts by paying their bills on time.

35 discounts that customers receive at the end of the “benefit period” vary from none to some. Those that continue to offer discounts typically replace them with lower discounts than they had made in their introductory offers.

The rationale for the 50/50 weighting adopted here is as follows:

• Data from the Australian Energy Market Operator shows that the typical annual residential switching rate (after excluding new homes in Victoria) is around 15- 20% per year. • Assuming this switching rate applies consistently to all consumers, the typical rate at which customers switch suppliers is every 4 to 6.7 years (the inverse of the annual switching rate). • This is well beyond the duration of the benefit period of all discounts. Many, probably most, customers will therefore be receiving none or at most some reduced discount (if they were always receiving the full introductory discounts why would retailers make them conditional and restrict them to defined “benefit periods”). • On the other hand, some retailers may not withdraw all discounts at the end of the benefit period and some customers may obtain cheaper offers from their existing retailers without switching.

Balancing these counter-vailing factors, the 50/50 assumption of the proportion of customers that are on fully discounted Market Offers versus Market Offers without conditional discounts is adopted.

It should be noted that in its Residential Price Trends Report, the AEMC assumes that all customers on Market Offers are on their retailer’s cheapest Market Offers. This is likely to underestimate customers’ bills and their average prices as quantified later in this section and verified in the analysis of the bill sample in Section 7.7

7 The representative customers’ bill applying the AEMC’s approach for offers valid on 14 May 2017 results in an annual bill for a 4 MWh per annum representative customer on a flat or block tariff of $1,100 before GST per year. This compares to the calculation using our approach which estimates the bill to be $1,326 before GST - 21% higher than the AEMC’s estimate for market offers in Victoria in the year to June 2017.

3.1.2 Estimating the wholesale charge

The Department commissioned Jacobs to provide monthly estimates of the wholesale prices applicable to residential customers in each of the four regions of the NEM. The issue is to decide how to use these monthly prices in estimating wholesale prices for the representative customer. There are many plausible ways that this could be done.

One approach would be simply to use Jacobs’ May wholesale price estimate. However the retail offers that are analysed in this report will remain valid for some future period before they will be changed. And, many of the offers currently available were introduced to the market some time previously.

A reasonable estimate of wholesale charges at any point in time should therefore take account of when the retail offers were introduced to the market and how long they are likely to remain in the market in future. This allows for a balanced estimate of the wholesale price for the duration of that retailer offer, to be established.

Two pieces of information are needed to make a plausible estimate of wholesale prices using Jacobs’ base monthly price data:

• firstly, how long the offers have been in the retail market when the market is assessed; and • second, how long offers currently in the market can reasonably be expected to remain in the market for.

On the first issue, this report analysed the retail market in May 2017. To estimate how long offers have been in the retail market, we analysed the tenure of the market offers to residential customers that were listed on Switchon, the Government’s price comparison website, at the end of 30 May 2017. Figure 13 below shows the offer duration (on the Y-axis) i.e. how long ago (from 30 May 2017) the offers were introduced to the market, for each of the 1017 market offers available on 30 May 2017 (on the x-axis).

Figure 13. Offer duration, all market offers, residential customers at 30 May 2017

500" 450" 400" 350" 300" 250" 200" 150" 100"

How$long$ago$the$oﬀer$was$introduced$(days)$$ 50" 0" 1" 41" 81" 121" 161" 201" 241" 281" 321" 361" 401" 441" 481" 521" 561" 601" 641" 681" 721" 761" 801" 841" 881" 921" 961" 1001" Market$oﬀer$$

As at 30 May 2017 the median offer had been introduced to the market 135 days earlier (4.5 months). Three quarters of all offers had been introduced to the market just under 5 months ago. From this information we chose an offer duration of 5 months, in other words that for the purpose of analysing wholesale prices we assume that the market offer for the representative customer was first made available to the market five months back from the end of 30 May.

Turning now to the second issue (how long offers can reasonably be expected to remain valid for) the duration of offers in the market was assessed by examining all market offers available to residential customers in December 2016. This date was chosen in recognition of the traditional re-pricing that occurs in Victoria around the start of the year. The analysis at the end of December showed a median duration of just over 10 months and a 75th percentile duration of just over 11 months. In other words, of the offers in the market at the end of December 2016, 75 percent had been introduced to the market at most 11 months previously.

Figure 13 shows that offers are increasingly being updated more frequently than the traditional annual re-pricing8. We presume this reflects retailers’ response to wholesale price changes i.e. they are changing (raising) their retail offers more frequently than they otherwise would have, in recognition of the recent increases in wholesale prices. Erring on the side of caution, we nevertheless assume that market offers will typically endure for 10 months before they are revised. This is a conservative estimate: it may well be that the effective duration is somewhat shorter. Certainly it is reasonable to expect offer duration to be shorter in future than it has been in the past.

Bringing the two strands of the analysis together (i.e. when offers for the representative customer were introduced to the market and how long these offers will endure in the market for) means that the appropriate estimate of wholesale prices in May 2017 is established by taking the average of Jacobs’ monthly wholesale price estimates for 10 months starting from 1 January 2017 (5 months back from 30 May 2017). This gives a wholesale price in Victoria of $66/MWh. The prices in New South Wales, Queensland and South Australia, established using the same approach are $77/MWh, $88/MWh and $146/MWh respectively.

It is important to be clear that this analysis does not suggest that wholesale prices will not be appreciably higher in future than they are today. Rather, the analysis reflects the assumption – based on the observed offer duration data – that retailers will reprice their offers to reflect future changes in wholesale prices. If the study in this Report was repeated in, say, October 2017, it would result in a higher wholesale price estimate but also a higher retail price estimate. The same principle would apply if the study was done at some earlier date, say December 2016, at which point it will have shown lower wholesale and retail prices.

Finally an important result in this study – the retailers’ charge for its services - is somewhat sensitive to wholesale price assumptions, but this charge is far more sensitive to estimates of the customers’ bills. For example, if the Victoria wholesale

8 This is also implicit in the December 2016 data in which the median is 10 months and even 75 percent of all offers were in the market for less than a year. In other words, even in 2016 many retailers were re-pricing offers that they might have first established in January.

39 price was assumed to 10% higher than the estimate used here (i.e. $72.6/MWh rather than $66/MWh) the retailers’ charge for its services would reduce from $489 per customer per year to $462 per customer. By contrast if retail bills were assumed to be 10% lower than the $1,326 per year calculated in this report, the retailers’ charge for its services would reduce from $489 to just $356 per year

3.1.3 Estimating network charge

The network charge in each distribution region is based on the relevant 2017 rates specified in the published tariff sheets of each distribution network service provider, for two-part (flat rates) and block tariffs for residential customers.

For small business customers the weighted average of the network charges on time of use and flat rate is calculated with the weighting set the proportion on time of use versus flat tariffs.

The weighted average Victorian network charge weights the network charge for each distributor based on the number of customers that that retailer serves compared to the total number of household customers. These network tariff calculations are performed in MarkIntell (Insight).

3.1.4 Estimating environmental charges

The Environmental charge is the sum of separate components for the Federal Small Scale Certificate Scheme and the Federal Large Scale Renewable Energy Target Scheme, the Victorian Premium Feed-in Tariff, and the Victorian Energy Efficiency Target.

Federal Small Scale Scheme This the product of the Small Scale Technology Percentage (as published by the Clean Energy Regulator) and the Small Scale Technology Certificate price which is assumed to be $40 per certificate.

Federal Large Scale Renewable Energy Target

This is the product of the Renewable Power Percentage (RPP) and the price of Large Scale Generation Certificates (LGCs). The price of LGCs is assumed to be $70 per certificate representing a $10 per certificate discount to the spot retail price at the time of writing. We suggest this discount reflects the likely average wholesale price for

LGC’s paid by retailers. The federal environmental charge calculations are performed in MarkIntell.

Victoria Energy Efficiency Target

This is estimated based on the Victoria Government’s 2017 Victorian Energy Efficiency Certificate (VEEC) liability priced at $16 per certificate based the average of the previous 12 months’ prices, published by Green Energy Trading, and assuming the cost is spread evenly over the 37 TWh distributed by Victoria’s distributors each year.

Premium feed-in tariffs

This is based on 2016 data of aggregate eligible feed-in volume of 120 GWh priced at $600 per MWh and spread evenly over the 37 TWh distributed by Victoria’s distributors each year.

3.1.5 Metering charges

This is the based on the prices in distributors’ tariff sheets for single element meters. weighted by the number of customers in each distribution area.

3.1.6 GST

This is calculated at the statutory rate on the sum of all the pre-GST elements of the bill.

3.2 Results

Figure 14 shows the breakdown of the residential electricity bill in Victoria based on the assumptions described in the previous sub-section. Figure 17 shows the breakdown for small business customers on the same assumptions.

Figure 14. Residential bill disaggregation based market offers, 4 MWh p.a.

RepresentaIve!annual!household!electricity!bill!! $1,600! !$133!! $1,400! !$18!! !$55!! !$89!! $1,200! !$266!!

$1,000! !$409!!

$800!

$600! !$489!!

$400!

$200!

$0! Retailer's!!!!!!!!!!!!!!!!!Network!!!!!!!!!!!Wholesale!!!!!!!!!!!!!!Metering!!!!!!!!!!!!! Federal! Victoria! GST! charge! charge! charge! charge! environmental! environmental!

Figure 15. Small business bill disaggregation, 10 MWh p.a.

RepresentaJve!annual!small!business!electricity!bill!! $3,500!

$3,000! !$268!! !$45!! !$137!! !$89!! $2,500! !$660!!

$2,000! !$1,098!!

$1,500!

$1,000! !$647!!

$500!

$0! Retailer's!!!!!!!!!!!!!!!!!Network!!!!!!!!!!!Wholesale!!!!!!!!!!!!!!Metering!!!!!!!!!!!!! Federal! Victoria! GST! charge! charge! charge! charge! environmental! environmental!

4 Inter-state and international comparison of prices and retailer charges in residential offers

This section extends the analysis of Victorian prices and retailer charges in residential offers by comparing them to those in other contestable markets in Australia and then comparing the Australian markets to those in Europe based on the analysis in the Agency for the Cooperation of Energy Regulators’ (ACER)/ Council of European Energy Regulators (CEER) 2016 report9.

ACER’s analysis assumes a 3.5 MWh per year customer. As explained earlier, we assumed a 4 MWh per year customer in our Victoria analysis. Considering the effect of high fixed charges in Australia this differences means that the comparison will understate Australian average prices (and retailer charges) relative to those in Europe. For example, on 11 May 2017, the median pre-tax price in market offers for a 4 MWh per year customer (assuming all discounts are met) is 28.2 cents per kWh. But for a 3.5 MWh customer the median price rises to 30.4 cents per kWh. On this basis, this international comparison should be considered favourable to Australia i.e. that it understates Australian prices and retailer charges, relative to the international comparators.

The comparison with Europe has used current market exchange rates (0.67 Euros per Australian dollar). This is about comparable to purchasing power parity exchange rates for western European countries. For some of the eastern and southern European countries PPP rates of exchange are lower so the comparison of prices and retailer charges with theirs, is more favourable to Australia. This is explore further in Part B.

The comparison distinguishes the pre-tax and taxation components. In several European countries with high electricity prices – for example Germany, Denmark,. Italy and Portugal - high prices are explained by taxes not by the prices charged by the electricity industry itself.

9 Agency for the Cooperation of Energy Regulators and Council of European Energy Regulators, November 2016. “Market Monitoring Report 2015 - ELECTRICITY AND GAS RETAIL MARKETS”. p.43

4.1 Prices

Figure 16 compares residential electricity prices before and after taxes in the contestable retail markets in Australia with the residential electricity prices in retail markets in Europe based on the methodology described in the previous section. It should be noted that the prices charged by the dominant retailers in New South Wales, South Australia and Queensland will increase from July and this will further extend the gap between the pre-tax prices in these states and those in Europe.

Figure 16. Comparison of residential electricity prices (before and after tax) (Australian cents per kWh) (May 2017 prices in Australia, 2015 prices in European countries)

!60!! Retail!prices!pre"tax! Taxes!!

!50!!

!40!!

!30!!

!20!!

!10!! Retail!price!(Australian!cents!per!kWh)! !"!!!!

Italy! Spain! France! Victoria! Ireland! Holland! Belgium!Austria!Greece!Sweden! Poland!Finland! Norway! Estonia! Germany! Slovakia! Slovenia! Denmark! Portugal!Hungary! Lithuania! EU!Average! Queensland!Great!Britain! Luxumbourg! South!Australia! New!South!Wales!

4.2 Retailer charges

Figure 17 compares the retailers’ charge to residential customers for the representative customers in Victoria compared to the other deregulated retail markets in Australia. However as noted above prices in these other states are shortly to rise and this will affect the comparison (as will future price changes in Victoria of course).

Figure 17. Inter-state comparison of retailer charges to residential customers

!$500!!

!$450!!

!$400!!

!$350!!

!$300!!

!$250!!

!$200!!

!$150!!

!$100!! Retailer(charges(($(per(customer(per(year)(

!$50!!

!$#!!!! NSW! SA! QLD! VIC!

Figure 18 extends the analysis by comparing retailer charges to residential customers in the retail markets in Australia to the outcomes in retail markets (mostly) in various European countries. Clearly retailers’ charges for their services in Australia, but particularly Victoria are far higher than those estimated by ACER for the various European countries.

Figure 18. Comparison of retailer charges to residential customers (cents per kWh)

14.0%

12.0%

10.0%

8.0%

6.0%

4.0%

Part B: Analysis of a sample of household electricity bills

5 Sample data description

This section describes the sample of household electricity bills that are analysed. It begins with a description of the data source and then of relevant aspects of the data. It ends with information on the estimated annual electricity bills based on the bill data.

Newgate collected customer bill information including copies of customers’ bills. Newgate captured the data in a database and undertook preliminary quality control. The Department undertook further quality control. We tested various data entries in the course of this analysis. The final dataset covered 686 unique residential electricity bills. The data from this that is used in this analysis includes:

• Retailer and network service provider identity for each bill • All relevant variable and fixed rates; • Solar feed-in rates (after excluding the statutory entitlements, where applicable); • All relevant discounts; • All relevant consumption amounts.

Figure 19 summarises the sample bills by distribution region and tariff type. All tariffs have daily charges (cents per day) and in addition:

• Flat tariffs have single consumption rates that apply at all times; • Flexible tariffs have peak, shoulder and off-peak rates and may also have block structures for peak rates; • Multiflat tariffs are block tariffs typically with at least two block rates; • Multi-tou tariffs contain a combination of block rates and time of use structures;

• Tou5 is a time of use tariff with oo-peak rates form 11pm to 7am on weekdays and peak rates at other times.

Figure 19. Distribution by tariff type and network service provider

Figure 20 summarises the sample bill data by tariff type and retailer. The share of retailers in this sample per retailer accords approximately with the known market share of the population, with the possible exception of Simply Energy which is over- represented in the sample relative to the population. Flat tariffs dominate for all retailers except Momentum whose dominant tariff type is multiflat.

Figure 20. Distribution of tariff type by retailer

Consumption The density distribution of consumption is an important feature of the sample. Figure 21 shows the density distribution of consumption of bills in the sample (the Y axis shows the percentage of the sample set corresponding to the consumption value on the X-axis). The top distribution in the figure shows the distribution of all bills in the sample. The lower distributions breaks the sample into three consumption clusters (low, medium and high) and shows the distribution of each on the same axis.10

The distributions in this figure, but particularly the density distribution of the aggregate dataset shows the skew distribution of consumption in the sample. This skew is also seen in consumption data in the population.11 This is an important characteristic of the data that needs to be taken into account. The skew means that the median (the middle value) is less than the mean (the average of all values). This is an important because the selection of the “typical” customer is often based on the median. Since the median is less than the mean, the “typical” customer consumes less than the average and, assuming unchanged prices, the typical customer’s bill will be less than the average bill. As shown later, the gap between median and average is quite large.

Figure 21. Density distribution of daily consumption by cluster and in aggregate

10 The clusters were determined in a R. package that defines clusters that have the minimum intra-cluster variance and maximum inter-cluster variance. 11 Ausnet Services, referenced earlier, has published a distribution of the consumption of small customers on its network and it shows the same skew distribution.

Retailer market share in sample versus population

The retailer market share in the sample compared to their known shares in the population at the end of 2016 is shown in Table 11 below. The table shows a reasonably close correspondence in the sample compared to the population (based on the customer numbers published by the Essential Services Commission in December 2016), although the larger retailers tend to be slightly under-represented in the sample and conversely the smaller retailers mostly slightly over-represented.

Table 11. Retailer market share in sample compared to population

Market' Market' share'*' share'*' sample population agl 19.1% 20.8% origin 17.3% 18.2% ea 14.6% 19.6% simply 12.5% 9.2% redenergy+lumo 12.5% 17.4% momentum 6.7% 2.7% powershop 5.0% 2.2% dodo 2.6% 2.3% alinta 2.3% 3.2% click 2.2% 1.4% globird 1.7% 0.2% powerdirect 1.5% 1.6% pacifichydro 0.7% 0.0% sumo 0.6% 0.4% opg 0.3% 0.3% peopleenergy 0.1% 0.4% next 0.1% 0.0% .

Fixed charge versus total charge

Figure 22 below charts the annual fixed charge versus the total annual bills for all bills in the sample. It shows a wide range (from just under $200 per year to just under $600 per year) and that fixed charges are a significant part of the bill even for high consumption customers (for whom fixed charges are relatively the smallest part of the bill).

Figure 22. Fixed charges versus total bill ($ per year)

Figure 23, analyses the difference between the fixed charges in the retail bill (after all relevant discounts) and compares it to the fixed network charge and metering charge and the difference between the fixed retail and sum of fixed network and metering. The pattern and quantum in this chart is consistent with the pattern and quantum of this analysis based on offers in the market (see Figure 11), rather actual bills.

Figure 23. Annual fixed retail, fixed network and metering charges

Retail! Network! Metering! Retail#network#metering! !$450!!

!$400!!

!$350!!

!$300!!

!$250!!

!$200!!

!$150!!

Fixed&charge&($&per&year)& !$100!!

!$50!!

!$#!!!! ausnet! ci2power! jemena! powercor! united!energy!

Discounts

Figure 24 shows histograms that describe the distribution of discounts by network service provider area. For the purpose of these histograms, discounts on usage were restated as a percentage discount on the total bill to ensure that discounts on usage and discounts on the total bill can be placed on a common footing. The charts show that in all distribution regions, discounts are common.

Figure 24. Distribution of discounts by network service provider

Figure 25 distinguishes the discounts on total bills from discounts applied to usage usage charges. In both cases, the discount rates cover a wide range.

Figure 25. Description of discounts

To understand other important aspects of discounts, Table 12 provides summary information distinguishing between discounts on usage and total bills and discounts that are conditional or unconditional. The most widely used discount (on about half of all bills in the sample) are conditional discounts on usage, followed by conditional discounts on the total bill and then unconditional discounts on usage. A few bills had discounts on both usage and the total bill that were conditional/unconditional and/or unconditional and conditional. Around 1/7th of all bills had no discount at all.

Table 12. Summary information of discounts by type and incidence

Number)of) 1st) 3rd) Discount)on Discount)type bills Min. Quartile. Median Mean Quartile. Max. energy unconditional 113 2% 3% 8% 14% 28% 42% energy conditional 314 1% 26% 28% 27% 35% 55% total)bill conditional 171 2% 10% 19% 18% 24% 40% total)bill unconditional 52 3% 10% 28% 22% 30% 33% energy)&)energy conditional)and)unconditional 57 6% 28% 30% 29% 34% 63% total)bill)&)energy unconditional)and)conditional) 9 6% 12% 30% 24% 34% 38% total)bill)and)total)bill) unconditional)and)conditional) 1 16% 16% 16% 16% 16% 16% No)discount No)discount 101 0 0 0 0 0 0

This picture is broadly consistent with the frequency and incidence of discounts in offers, as discussed in Part A.

While information on the discounts themselves is useful showing the effect they have on bills is of course where they can really be measured. The measure here is how competitive rates are after all discounts have been deducted. This competitiveness can be measured by establishing how much customers would save if they switched from their offer to the least expensive offer in the market. This is shown in Figure 26 which distinguishes those bills with discounts in four categories (none, 0-10%, 10%-30% and > 30 %). This group of charts shows that savings vary considerably across the clusters and within the clusters, as measured by the low R-squared scores of the lines of best fit of the data in each cluster. The charts show that some of the bills that have no discount are hard to beat in the market. Only for those bills with more than a 30% discount is there a relatively consistent upper limit (at around $300) of the savings that customers on those bills might expect from switching to the least expensive offer in the market.

Figure 26. Switching saving: impact of discount rate

Solar feed-in rates

Only the retailer component of feed-in tariffs is included in this analysis. The statutory (“Premium Feed-in-Tariff”) component (if applicable) will reduce the bill to the relevant customer but does not affect the analysis of retailers’ charges for their services. Figure 27 describes retailers’ solar feed-in rates (stated in $/kWh) of those in bills in the sample with solar feed-in tariffs.

Figure 27. Description of retailers’ solar feed-in rates

Estimated annual bills

An important part of this study is the calculation of the estimated annual bill based on the bills in this study. The bills, typically for 30 or 90 day billing periods are annualised assuming that the consumption, and if applicable the consumption pattern, is representative of the annual consumption and pattern of that customer. In practice this might not always be the case but there is unlikely to be any significant asymmetric

54 error in this annualisation assumption12. Even if this were not the case, any asymmetric error is likely to be small. By testing different annualisation assumptions13 we found that even reasonably large asymmetric annualisation errors do not affect the main observations in this Report.

Concession payments are not included in this analysis. There are 125 bills in the sample that receive some form of concession payment. This does not affect the calculation of the retailers’ charges for their services, but will affect the analysis of the bill and price for customers entitled to the concession.

Estimates of annual charges and customer prices include GST (in recognition of the fact that GST is rarely deductible to households). Table 13 (ordered by number of bills in the sample) presents summary statistics of the estimated annual charges for each retailer in the sample.

Table 13. Summary statistics of estimated annual bills ($ per customer per year)

retailer count Min. 1st$Qu. Median Mean 3rd$Qu. Max. agl 131 $""""""""""!531 $""""""""""!935 $"""""""!1,331 $"""""""!1,518 $"""""""!1,743 $"""""""!6,801 origin 119 $""""""""""!211 $""""""""""!906 $"""""""!1,267 $"""""""!1,397 $"""""""!1,666 $"""""""!4,523 ea 100 $""""""""""!368 $""""""""""!929 $"""""""!1,269 $"""""""!1,414 $"""""""!1,709 $"""""""!5,470 simply 86 $""""""""""!427 $""""""""""!791 $"""""""!1,049 $"""""""!1,220 $"""""""!1,443 $"""""""!4,433 redenergy 51 $""""""""""!320 $""""""""""!723 $""""""""""!950 $"""""""!1,185 $"""""""!1,518 $"""""""!3,118 momentum 46 $""""""""""!488 $""""""""""!816 $"""""""!1,115 $"""""""!1,244 $"""""""!1,542 $"""""""!2,788 lumo 35 $""""""""""!396 $""""""""""!812 $"""""""!1,208 $"""""""!1,392 $"""""""!1,640 $"""""""!3,539 powershop 34 $""""""""""!480 $""""""""""!824 $""""""""""!955 $"""""""!1,167 $"""""""!1,418 $"""""""!2,975 dodo 18 $""""""""""!434 $""""""""""!785 $"""""""!1,049 $"""""""!1,167 $"""""""!1,365 $"""""""!2,703 alinta 16 $""""""""""!519 $""""""""""!999 $"""""""!1,228 $"""""""!1,436 $"""""""!1,743 $"""""""!3,182 click 15 $""""""""""!574 $""""""""""!991 $"""""""!1,214 $"""""""!1,361 $"""""""!1,601 $"""""""!3,045 globird 12 !$###############!3 $""""""""""!396 $""""""""""!756 $""""""""""!700 $""""""""""!943 $"""""""!1,354 powerdirect 10 $"""""""!1,133 $"""""""!1,409 $"""""""!1,645 $"""""""!1,785 $"""""""!2,167 $"""""""!2,911 pacifichydro 5 $""""""""""!481 $""""""""""!589 $""""""""""!742 $"""""""!1,023 $""""""""""!787 $"""""""!2,518 sumo 4 $"""""""!1,058 $"""""""!1,094 $"""""""!1,302 $"""""""!1,549 $"""""""!1,756 $"""""""!2,534 opg 2 $""""""""""!988 $"""""""!1,090 $"""""""!1,191 $"""""""!1,191 $"""""""!1,293 $"""""""!1,395 peopleenergy 1 $"""""""!1,986 $"""""""!1,986 $"""""""!1,986 $"""""""!1,986 $"""""""!1,986 $"""""""!1,986 next 1 $""""""""""!581 $""""""""""!581 $""""""""""!581 $""""""""""!581 $""""""""""!581 $""""""""""!581

12 Net system load profile data does not suggest that consumption in the period December to April during which most consumption in this sample occurred is meaningfully different from average annual consumption. 13 Assuming the consumption in the annual bill was 10% above or below average for the bill.

Figure 28 shows the distribution of the annual bill by distribution region. A few observations stand out:

• Customers bills vary over a wide range; • In all cases the distribution is skewed (the median is less than the mean in all but one case);

• The median (and mean) vary by distribution zone in a way that is consistent with the analysis of offers in the Part A Report (charges are higher in those distribution areas – Ausnet and Powercor – where network charges are higher).

Figure 28. Distribution of the estimated annual bills ($ per customer per year) in each distribution zone

The summary statistics presented in Table 14 takes the estimated annual bills and divides them by the estimated annual consumption to yield the estimated annual prices (cents per kWh). As expected, average prices vary widely, but again the differences between the median and average prices reflect the skew in the distribution of consumption in the sample.

Table 14. Summary statistics of estimated annual prices (cents per kWh)

Number+of+ retailer bills Min. 1st+Qu. Median Mean 3rd+Qu. Max. agl 131 17 27 32 37 40 180 origin 119 14 28 33 37 39 103 ea 100 19 29 34 36 40 77 simply 86 16 24 28 31 36 65 redenergy 51 16 28 35 36 40 82 momentum 46 13 24 29 30 34 62 lumo 35 13 26 34 34 40 61 powershop 34 23 28 32 33 35 58 dodo 18 21 29 35 37 40 83 alinta 16 25 31 35 48 45 212 click 15 25 27 31 33 38 49 globird 12 0 28 30 33 40 58 powerdirect 10 21 23 29 27 30 32 pacifichydro 5 18 24 27 29 34 42 sumo 4 29 30 31 31 31 32 opg 2 29 32 34 34 37 40 peopleenergy 1 30 30 30 30 30 30 next 1 35 35 35 35 35 35

It merits observation here that the average price for the Big Three retailers here – 34 cents per kWh before GST is 22% higher than the Australian Energy Markets Commission calculates in its latest report on the competitiveness of the retail electricity industry.14

14 Australian Energy Markets Commission, July 2017. “2017 AEMC Retail Energy Competition Review”. Page 112.

6 Retailers’ charges in the sample bills

This section breaks down (disaggregates) the bills in the sample into their component parts, distinguishing between the wholesale charge (the charge for electricity production), the network charge (the charge for transporting electricity over the transmission and distribution networks), the regulated charge for smart meters, the charge for federal and Victorian Government environmental programs, and finally the retailers’ charges for their service of retailing electricity. The focus here is particularly on understanding the retailers’ charge and how it compares to the other parts that make up the customers’ bill. The section focusses on developing point estimates of the retailers’ charges to representative households, comparing retailers’ charges relative to network and wholesale charges across the sample, and finally showing how the retailer charges compares within and between retailers.

All 686 bills including 75 with controlled load rates and 102 with solar feed-in and three with both are included in the analysis. For the bills with solar feed-in we assume that the opportunity cost to retailers of purchasing fed-in electricity is the wholesale price of electricity. So retailers that pay feed-in rates lower than the assumed wholesale price will improve their margins from those customers (by the difference between the wholesale price and the feed-in rate they pay). Conversely retailers that pay a price for fed-in electricity that is higher than the assumed wholesale price will lose margin on those customers based on the difference between the wholesale price and feed-in rates they pay.

The wholesale and environmental prices and charges are established as described in Part A. The application of the wholesale price methodology described in Part A results in a wholesale charge of $61.6/MWh for electricity billed in this December 2016 to April 2017 period.15

15 Wholesale electricity prices have increased significantly in Victoria over the last year, though the measure of this depends on which contracts are examined or how spot prices are analysed. The approach in this Report to setting a wholesale price in the analysis of customer bills, as described in Part A, adopts the typical assumption that retailers hedge their prices in advance of their retail sales. This means that wholesale price estimates lag the current day prices in contract markets (which hedge future

Wholesale price assumptions sometimes attract an undue amount of attention considering that in the analysis of retailer charges it is not the most important factor. Specifically, if we assume a 4 MWh per year household, a $1 change in the assumed wholesale price results in a $4 change in the estimate of the retailers’ charge for its services. So changing the assumed wholesale price from say $50 per MWh to $70 per MWh (a 40% increase) will change the estimate of the retailers’ charge by $80 per customer per year. This is less than a 16% increase in the estimated retailer’s charge. This is represented graphically in Figure 29 below:

prices). Wholesale prices contracts for the period December 2017 to April 2018 are very much higher – perhaps even twice as high – as they have been for the same period a year earlier (i.e. the time that the bills in this study fall due). Clearly the calculation of retailers’ charges for the period from December 2017 to April 2018 should properly take account of the reasonably estimated wholesale prices at this time, and also the retail prices that will apply that time. From this it should be clear that the relevant issue in wholesale price estimates for retailer charge calculations is to ensure that the relevant retail prices are measured at the same time to which those wholesale prices apply. We do not yet know what prices retailers intend to charge for sales in the period from December 2017 to April 2018. At the time of writing this report, retailers in Victoria as elsewhere in Australia have been increasing their prices by large amounts. If retail prices do not rise later this year, this will mean that retailers’ charge later this year will become meaningfully lower than estimated in this report. A study of retailers’ charges in future should use the information that will become available on retail prices in future, along with the much higher estimates of wholesale prices applicable to that time, in the calculation of retailer charges at that time. Wholesale price assumptions are relevant also in the analysis of the relative competitive position of the retailers. The retailers with the largest number of customers in Victoria typically also own most of the electricity production in Victoria, though the proportion of own production does vary amongst the largest retailers. For the smaller retailers, almost none of whom produce any of the electricity that they sell, higher wholesale prices translate into small retailer charges without any offsetting gain from higher wholesale margins. This is an important difference and should be considered in the analysis of Victoria’s retail markets, but it is beyond the scope of this study.

Figure 29. Retailers' charge ($ per year) versus wholesale price ($/MWh)

$500$

$450$

$400$

$350$

$300$

$250$ Retailers'*charge*($*/*customer*/*year)*

$200$ $40$ $43$ $45$ $48$ $50$ $53$ $55$ $58$ $60$ $63$ $65$ $68$ $70$ $73$ $75$ $78$ Wholesale*price*($/MWh)

Consequently, the observation (discussed later) that retailers’ charge for their services are generally greater than the wholesale charge is robust to the estimate of the wholesale prices that is used in this Report.

The calculation of environmental and metering charges in this analysis are set out in the Part A analysis.16 The network charge is established for each bill based on the network tariff that matches precisely or if not most closely, the structure of the retail tariff and where applicable includes controlled load rates. This is automated in the analytical tool – SwitchGov – and all bills and their constituent elements are priced in SwitchGov, which is described in the Appendix.

Two important features merit discussion in understanding disaggregated bills:

• Firstly, as shown earlier, the distribution of consumption is leftward skewed i.e. there are more bills in the sample whose annual consumption is less than the average consumption, than there are whose consumption is more than the average (the mean is higher than the median).

• Secondly, also as discussed earlier, fixed charges form a significant part of the bill for all but the very largest customers. The effect of the high fixed charges

16 These are $109, $85, $85, $85 and $60 per year in Ausnet, Citipower, Jemena, Powercor and United Energy.

means that customers that purchase small amounts of electricity pay higher prices – per kWh – than those that purchase large amounts of electricity, even if their total annual bill is lower. Consequently the retailers’ charge for its services forms a larger part of the bill for smaller customers, even if in absolute terms the retailers’ charge is smaller than it is for large customers.

Retailer charge estimates for representative customers

Disaggregation of annual bills using the average annual bill and median annual bill is presented in Table 15 below. In this table, the second and third columns show the disaggregation on average and for the median bill. The third and fourth columns present this information in cents per kWh, by dividing each element of the bill by the Annual grid purchases. The fifth and sixth columns present the elements of the bill as a percentage of the total bill (before GST).

Table 15. Analysis of average and median bills in the sample

Annual&bill&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&Percentage)of)bill) Annual&bill&($&per&year) (cents'per'kWh) before&GST

Average Median Average Median Average Median Retail'(before'GST) $""""" 1,247! $""" !1,094 !!!! !29.0 !!!! !37.9 Network((before(GST) $"""""""" 436! $"""""" !394 !!!! !10.1 !!!! !13.7 34.9% 36.0% Wholesale((before(GST) $"""""""" 265! $"""""" !178 !!!!!! !6.2 !!!!!! !6.2 21.2% 16.2%

Federal'environmental'(before'GST) $"""""""""" 55! $"""""""" !35 !!!!!! !1.3 !!!!!! !1.2 4.4% 3.2% Victoria(environmental((before(GST) $"""""""""" 16! $"""""""" !11 !!!!!! !0.4 !!!!!! !0.4 1.3% 1.0% Metering((before(GST) $"""""""""" 87! $"""""" !109 !!!!!! !2.0 !!!!!! !3.8 6.9% 10.0% Retailers'*charge*(before*GST) $"""""""" 389! $"""""" !368 !!!!!! !9.1 !!!! !12.8 31.2% 33.6% GST $"""""""" 125! $"""""" !109 !!!!!! !2.9 !!!!!! !3.8 Annual&grid&purchases&&(kWh) !!!!!!! !4,295 !!!!! !2,884 !!!4,295 !!!2,884

Several observations stand out:

1. As expected, the volume of electricity purchased from the grid for the median customer is less than it is on average (this reflects the skewed consumption distribution). 2. Also as expected, the median retailers’ charge does not decrease in proportion to the consumption (this reflects the effect of the large fixed charges in typical retail bills). Specifically, the retailers’ charge for the average and median is much the same, while the median customer buys much less electricity than the

average. This means that lower consumption customers pay much more per kWh for the retailers’ services, than higher consumption customers. This also translates into significantly higher prices per kWh purchased for the median customer compared to the average. 3. For both the average and median bill, the retailers’ charge and network charge are approximately comparable as a percentage of the total bill.

The median bill shown in Table 15 has a consumption of 2,884 kWh, this is far below the median consumption in the sample of 3,558 kWh per year and results in an estimate of annual prices that are far above average prices. The retailer charge estimate for this median customer is very different to the retailer charge for the customer either side of the median.

The average bill is a better choice for characterisation of representative customer, but is susceptible to the criticism that with a skewed consumption distribution it can not be accepted as an estimate of the bill for a customer whose annual consumption is typical of the population.

To characterise a representative customer, we have selected those bills whose consumption is in the range from 3,850 kWh per year to 4,150 kWh per year (3.75% either side of 4,000 kWh per year). In the sample there are 36 bills with annual consumption in this range and the average of their consumption is 4,007 kWh per year, close to the Victorian Government’s assumption of 4,026 kWh as the annual consumption of the typical residential customer. A waterfall chart of the disaggregated bill based on the average bill for customers in this range is shown in Figure 30.

The average bill in this sample is $1,389 (after GST), giving an average price of 34.7 cents per kWh. The average retailers’ charge for these customers is $423 per year, with a standard deviation of $167. This level of variability is an inevitable consequence of the high level of dispersion of the prices paid by customers in this consumption range and indeed in the whole sample. In properly understanding the retailers’ charge it is important to focus not just on the representative customer but also on the outcomes for all customers as shown in more detail in Figure 31, 32, 33 and Table 16.

Figure 30. Residential bill disaggregation based on average bill from sample (4 MWh)

Representativehousehold electricity%bill%from%sample% $1,600

$1,400 $126% $18% $55% $88% $1,200 $263%

$1,000 $415% $800

$600 $423% $400

$200

$0 Retailer's%%%%%%%%%%%%%%%%%Network%%%%%%%%%%%Wholesale%%%%%%%%%%%%%%Metering%%%%%%%%%%%%% Federal% Victoria% GST charge charge charge charge environmental environmental

Retailers’ charge relative to network and wholesale charges across the sample

Recognising the challenge of defining a central estimate for the sample, it is valuable to understand the outcomes for all bills in the sample. Figure 31 shows the difference between the retailers’ charge and the network charge for all bills in the sample. It shows that retailer charges are mostly higher than network charges except for customers in Ausnet Services’ and possibly also Powercor’s area of supply.

Figure 31. Histogram of the difference between network and retailers' charges

Figure 32 compares the retailers’ charges and wholesale (production) charges across the sample by showing the difference between the retailers’ charge and wholesale charge for all bills in the sample. This shows that for around three quarters of all customers in the sample, the retailers’ charge for their services in selling electricity is more than the charge for producing that electricity.

Figure 32. Histogram of the difference between wholesale and retailers' charges

Retailers’ charge analysed by distribution area

Figure 33 below presents histograms of the retailers’ charge by distribution region. It shows, as expected, a skew in the distribution of the retailers’ charge, originating in the skew in the consumption distribution. In these charts, for the purpose of presentational clarity, the scale has been limited to $1000 but there are several bills with annual retailers’ charge of $2,000 per year or more. Closer inspection of the charts also shows reasonable differences in the distribution of the retailers’ charges in each distribution area.

Figure 33. Histogram of retailers' charge by network service provider area

Retailers’ charges analysed by retailer

It also useful to know how the retailers’ charges for their services compare amongst the different retailers in the sample. Table 16 presents summary statistics on retailer charges (expressed as dollars per customer per year) calculated according to this methodology and ranked in the table in order of the number of bills supplied by each retailer in the sample.

Table 16. Summary statistics on retailer charges for their services ($ per customer per year)

Number+of+ retailer bills Min. 1st+Qu. Median Mean 3rd+Qu. Max. agl #############131 &$###########228 $###########322 $###########376 $###########468 $###########479 $########2,361 origin ##############119 &$###########146 $###########277 $###########380 $###########447 $###########552 $########1,401 ea ##############100 &$#############49# $###########242 $###########342 $###########389 $###########505 $########2,060 simply ################86 $#############11# $###########148 $###########223 $###########233 $###########262 $###########882 redenergy ################51 &$###########211 $###########232 $###########321 $###########324 $###########435 $###########744 momentum ################46 &$###########243 $###########225 $###########259 $###########257 $###########290 $###########544 lumo ################35 &$###########132 $###########248 $###########332 $###########383 $###########473 $########1,565 powershop ################34 $###########122 $###########240 $###########298 $###########378 $###########433 $########1,075 dodo ################18 &$###########138 $###########201 $###########280 $###########241 $###########329 $###########385 alinta ################16 $###########250 $###########348 $###########392 $###########477 $###########527 $###########902 click ################15 $###########191 $###########263 $###########430 $###########404 $###########500 $###########727 globird ################12 &$###########339 $###########110 $###########167 $###########118 $###########206 $###########318 powerdirect ################10 $###########146 $###########243 $###########323 $###########392 $###########389 $###########949 pacifichydro ##################5 $################9 $#############35# $#############44# $#############61# $###########105 $###########112 sumo ##################4 $###########170 $###########236 $###########348 $###########344 $###########455 $###########509 opg ##################2 $###########268 $###########289 $###########310 $###########310 $###########330 $###########351 peopleenergy##################1 $###########351 $###########351 $###########351 $###########351 $###########351 $###########351 next ##################1 $###########114 $###########114 $###########114 $###########114 $###########114 $###########114

In this table, the three largest retailers are shown to have median retailer charges in the range from $342 to $380 per year. The average retailers’ charge is around 10% higher than the median reflecting the skew in the distribution of consumption.

The table also shows that for several retailers the minimum retailer charge is negative. As discussed in further detail later, 19 out of the 686 bills have prices that are lower than any of the commonly available offers in the market at the time of those bills. Figure 34 displays the retailer charge calculated as a percentage of the total bill, for each retailer

Figure 34. Distribution of retailer charges as a percentage of total bill (before GST), by retailer

This figure shows that the retailers with the greatest number of customers also tend to have the highest retailer charge. In other words, in the sample the retailers with the largest market share are also those that charge more for their services than the retailers with the smallest market share. The gap between the established, popular retailers and the newer entrants is large. This is also evident in the savings that customers might obtain if they switched to the least expensive retailers. This is explored in further detail in the next Section.

Cross-country comparison of retailers’ charge

The comparison of the retailers’ charge (as in Part A – see Figure 18) in Victoria and other parts of Australia with the retailers’ charge in various European countries17, using

17 The European data is taken from Figure 21 of the ACER/CEER Report which provides an annual average from 2008 to 2015. The retailer charges in 2015 were higher in some countries most notably in Britain, compared to the 2008 to 2015 average, but the Report attributed this to the depreciation of the Pound after the Brexit vote and so this increase reflects primarily a currency effect. Using the 2015 measure, the retailers’ charge in Britain is a little below that for New South Wales. The retailers’ charge was also appreciably higher in 2015 than the average for 2008 to 2015 in Germany, Ireland, Belgium, Austria, Greece and France. The higher 2015 numbers narrow the gap to Victoria, but in all cases a significant gap still remains.

67 the retailers’ charge based on the 4 MWh cohort average in Australia, is shown in Figure 35 below.

Figure 35. Cross-country comparison of retailers' charge (cents per kWh) using sample of bills

12.0%

10.0%

8.0%

6.0%

4.0%

As described in Part A, the retailers’ charge for the European countries is taken from the Agency for the Cooperation of Energy Regulators (ACER) and Council of European Energy Regulators (CEER) Report18. That Report uses the same approach in this Report, of subtracting production and distribution costs from the estimate of the representative bill19. The bills for the European countries assume consumption of 3,500 kWh per year, whereas the Victorian prices assume 4,000 kWh20. Considering the effect of fixed charges, particularly in Victoria, the retailers’ charge in Victoria per kWh would be about one cent per kWh (about 10%) higher than shown in the Figure 31 had we assumed a consumption level of 3,500 kWh rather than 4,000 kWh per year. Similar

18 Agency for the Cooperation of Energy Regulators (ACER) and Council of European Energy Regulators (CEER), November 2016. “ACER Market Monitoring Report 2015 - ELECTRICITY AND GAS RETAIL MARKETS”. 19 Their Report refers to what we call the “retailers’ charge” as the “mark-up” or “gross profitability” 20 To be precise, for NSW, QLD and SA we use 4,800 kWh per year using exactly the same analysis for these states as in Part A since actual bill data is not available for these states.

68 increases would apply in the other Australian states had we also used a consumption level of 3,500 kWh for them.

It should also be noted that the estimates for New South Wales, Queensland and South Australia are for May 2017, before significant price rises in each of these states from late June 2017 onwards.

ACER/CEER, like us, needed to estimate the wholesale element of the bill in order to pull apart the wholesale from retailers’ charges. Their report explains that their methodology for estimating wholesale charges “is based on the assumption that suppliers are rational and apply a ‘close-to-optimal’ procurement strategy”21 . As such, the retailers’ charge estimates that they derived are higher than they would be had they made no such assumption22. By contrast, our estimate of wholesale price makes no such “close-to- optimal” assumption in estimating wholesale prices. As such the gap between Australian and European country retailers’ charge estimates are likely to be further understated for this reason.

It might be argued that the use of a market exchange rate in converting between the Euro and the Australian dollar undermines the validity of this comparison since it fails to take account of the relative purchasing power parity between Australia and the various European countries.23 This argument has limited value – the relative purchasing power based on the OECD’s PPP data shows that AUD/EURO market exchange rates are only overvalued a little (around 20% in some cases but often less) in comparison between Australia and the wealthy European countries. While the AUD has greater purchasing power in the cohort of less wealthy European countries, the gap is still not nearly large enough to close the gap in their retailers’ charges relative to

21 Ibid, p. 42. 22 Because the wholesale charge estimate is lower than it otherwise would be. 23 For example, in its response to a private submission by the author of this Report, the Independent Pricing and Regulatory Tribunal rejected an international comparison of retailers’ charges along the same lines as shown in Figure 34 for these reasons – see IPaRT 2016, “Review of the performance and competitiveness of the retail electricity market in NSW”. Sydney. Page 20. From 1 July 2015 to 30 June 2016

Victoria. It should be noted also be noted that ACER/CEER does not use PPP adjustments in their own cross-country comparisons.

Another way to compare retailers’ charges in Victoria with those in the European countries is to take the ratio of the retailers’ charge to the wholesale charge in each country, using the Euro prices for the European countries and Australian dollar prices for Australia. This takes away any need to convert from one currency to another. The result of this is shown in Figure 36 below. Consistent with the results in Figure 34, the retailers’ charge relative to the wholesale charge in Victoria for the representative customer is far higher than in the European countries (more than double the next highest, Germany).

Figure 36. Ratio of retailers' charge to wholesale charge

Ratio%of%Retail%/%Wholesale%charges 1.8%

1.6%

1.4%

1.2%

1.0%

0.8%

0.6%

0.4%

0.2%

7 Switching savings

It has long been suggested that customers can get a better deal by shopping around. This analysis tests this by quantifying how much customers in the sample could reduce their bills if they switched to the least expensive offers; or to the least expensive offer from their existing retailer. It also explores what savings would be if the customer switched to the second, third, fourth and up to the tenth cheapest offer.

In this analysis, the dataset of commonly available offers against which the bills were compared was established by scraping the websites of all commonly available offers from all of the relevant retailers. These data are available in MarkIntell (Offer) which is described in Appendix A. The comparison for each offer is based on the offers that were available in the same month as the date of each bill.

The comparison of the customers’ bill with what it otherwise would be on all commonly available offers from all licensed retailers, entails pricing all of the competing offers for the same parameters (annual consumption, consumption pattern if applicable, existence of solar feed-in, existence of controlled load) specific to that customer. Although in principle, customers can choose to change from one tariff structure to another, in practice few retailers allow this and so the comparator offers in each bill are restricted to offers with the same tariff structure.

The analysis involves first pricing (working out the charges) in each of the bills based on all the relevant details of each bill. Then secondly, for each bill, working out what the charge would be for that specific customer, on all other commonly available offers in the market at the time that the bill was due. These competing offers are then ranked to find the cheapest alternative market offer for each bill.

The rest of this section discusses in order the results of the analysis for switching to the least expensive offer in the market and then switching to the least expensive offer from the customers’ existing retailer.

7.1 Savings by switching to the least expensive offer in the market

Table 17 presents summary statistics of the estimated annual saving that customers in the sample would obtain if they switched to the least expensive offer.

Table 17. Summary statistics of estimated saving from switching to the lowest market offers ($ per customer per year)

Number+of+ retailer bills Min. 1st+Qu. Median Mean 3rd+Qu. Max. agl 131 %$'''''''''''395 $'''''''''''207 $'''''''''''295 $'''''''''''380 $'''''''''''412 $''''''''2,553 origin 119 %$'''''''''''205 $'''''''''''184 $'''''''''''296 $'''''''''''372 $'''''''''''494 $''''''''1,332 ea 100 %$'''''''''''''72' $'''''''''''166 $'''''''''''272 $'''''''''''333 $'''''''''''427 $''''''''2,131 simply 86 %$'''''''''''115 $'''''''''''''33' $'''''''''''''97' $'''''''''''124 $'''''''''''174 $''''''''1,052 redenergy 51 $'''''''''''''52' $'''''''''''155 $'''''''''''231 $'''''''''''278 $'''''''''''370 $'''''''''''689 momentum 46 $'''''''''''''12' $'''''''''''''87' $'''''''''''123 $'''''''''''140 $'''''''''''191 $'''''''''''395 lumo 35 $'''''''''''''11' $'''''''''''117 $'''''''''''252 $'''''''''''287 $'''''''''''362 $''''''''1,213 powershop 34 $''''''''''''''''5 $'''''''''''126 $'''''''''''162 $'''''''''''209 $'''''''''''241 $''''''''1,118 dodo 18 $'''''''''''''67' $'''''''''''114 $'''''''''''152 $'''''''''''193 $'''''''''''221 $'''''''''''486 alinta 16 $'''''''''''186 $'''''''''''233 $'''''''''''288 $'''''''''''377 $'''''''''''396 $'''''''''''842 click 15 $'''''''''''''19' $'''''''''''128 $'''''''''''266 $'''''''''''265 $'''''''''''349 $'''''''''''643 globird 12 $''''''''''''% $'''''''''''''27' $'''''''''''''65' $'''''''''''''71' $'''''''''''''90' $'''''''''''215 powerdirect 10 $''''''''''''''''9 $'''''''''''130 $'''''''''''161 $'''''''''''266 $'''''''''''243 $'''''''''''830 pacifichydro 5 %$'''''''''''''29' $''''''''''''% $'''''''''''''22' $'''''''''''''26' $'''''''''''''45' $'''''''''''''93' sumo 4 $'''''''''''''49' $'''''''''''117 $'''''''''''196 $'''''''''''231 $'''''''''''310 $'''''''''''483 opg 2 $'''''''''''204 $'''''''''''208 $'''''''''''212 $'''''''''''212 $'''''''''''216 $'''''''''''220 next 1 $'''''''''''''23' $'''''''''''''23' $'''''''''''''23' $'''''''''''''23' $'''''''''''''23' $'''''''''''''23' peopleenergy 1 $'''''''''''259 $'''''''''''259 $'''''''''''259 $'''''''''''259 $'''''''''''259 $'''''''''''259

The table shows, as expected, that the average saving across the sample is higher than the median saving. The three largest retailers offer a small number of their existing customers (5 in total) better deals than they could get from the commonly available offers in the market. But as discussed later in more detail, there is a strong correlation between the median saving from switching away from the current retailer and the number of bills of that retailer in the sample. In other words, those customers of the retailers with the largest market share also tends to have the greatest likelihood of find the largest savings by switching retailer. Figure 37 shows the distribution of the annual saving, ordered by network service provider area within which the customer is located. It shows a similar distribution in all areas.

Figure 37. Distribution of annual saving by selecting the cheapest offer, by network service provider area

Figure 38 shows the distribution of the switching saving as a percentage of the respective customers’ bill.

Figure 38. Distribution of saving (as a percentage of total bill) by selecting the cheapest offer, by network service provider area

Figure 39 shows the impact of controlled load tariffs on switching savings. It shows that the existence of a controlled load does not have a large impact on the savings available to those customers, relative to others without controlled load, if they switched to the least expensive offers available to them.

Figure 39. Impact of controlled load on saving (as a percentage of the total bill) by selecting the cheapest offer

Figure 40 repeats the analysis in Figure 39 but focusses on whether customers have solar rather than controlled load. It shows that that the median and range of savings (as a percentage of bills) is similar for households that have solar PV and those that don’t have solar PV.

Figure 40. Impact of solar on saving (as a percentage of the total bill) by selecting the cheapest offer

Figure 41 shows the savings expressed as an annual dollar amount and as a percentage of the total bill for the three consumption clusters described earlier in the section describing the data. This shows the wide range of savings (in dollars) in each cluster but also that the median saving rate in each cluster is similar at around 21%. This is alternatively stated as an average saving of $294 per customer.

Figure 41. Distribution of savings by cluster

7.2 Cluster analysis of saving

We identified three clusters to categorise the saving that customers would achieve by switching retailer. This is shown in Figure 42. The y-axis shows the savings that customers would obtain if they switched to the cheapest offer in the market. In the figure, the bills are collected into three clusters:

In addition to these three clusters as shown, there were 33 bills that had no saving or were cheaper than any offer in the market and there were 8 bills with savings of more than $1,000 per year.

Figure 42. Switching savings in three clusters

7.3 Impact of customers’ ability to identify the cheapest offer

The savings calculations here are worked out for each customer after having found the cheapest offer for that customer based on the profile of their consumption, their existing prices and whether they have controlled load and/or solar. All offers in the market were scanned to find the right offer for that customer at the date of their bill. This is a sophisticated and data intensive analysis that will be beyond all but the very best informed customers. Furthermore the tools needed for this sort of comparison are not publicly available. A question that therefore needs to be answered is how much customers might save if they were not able to find the cheapest offer available to them, but instead the second, third, fourth and up to tenth cheapest offer. This analysis was undertaken and the results are shown in Table 18 below (inclusive of the GST). The table shows, in rows, the savings available if the first to tenth cheapest offer for each customer was selected. The second column shows the number of bills with a switching saving greater than zero. The third column shows the total annual saving if all customers in the sample switched to the cheapest offer. The statistical summary that follows in the remaining columns then describes the information on the savings in the sample.

The table shows that there is a big gap between the saving that customers would obtain if they selected the cheapest offer, than the second cheapest. For example, the median

77 annual saving drops from $218 to $106 and the total saving for all bills in the sample drops from $190,009 per year to $60,0025 per year. The gap between the second and third, third and fourth and so on is smaller than between the first and second, but if customers were only able to identify the fifth cheapest offer applicable to them, the median saving would have more than halved from $218 to $108 per year. The conclusion from this is that while the retail market offers many of the customers in the sample the ability to reduce their bill by switching, in practice the savings they obtain depends strongly on their ability to find the cheapest offer applicable to them. In this regard it is interesting, for example, that an offer that frequently appeared as the least expensive offer is not identified in any of the commercial switching sites in Victoria.

Table 18. Relationship between switching saving and selection of alternative offers (inclusive of GST)

Sum)of) Number)of) saving) bills)with) across)all) Switching)to: saving bills Min. 1st)Qu. Median Mean 3rd)Qu. Max. Lowest)offer 647 $190,009 $1 $128 $218 $294 $369 $2,393 Second)lowest)offer 586 $133,316 $0 $87 $156 $228 $277 $2,190 Third)lowest)offer 509 $103,925 $0 $63 $125 $205 $251 $2,030 Fourth)lowest)offer 476 $94,950 $0 $54 $114 $199 $248 $2,002 Fifth)lowest)offer 442 $85,857 $0 $50 $108 $195 $251 $1,981 Sixth)lowest)offer 419 $79,519 $0 $45 $103 $190 $244 $1,916 Seventh)lowest)offer 401 $74,546 $0 $44 $100 $186 $237 $1,862 Eigth)lowest)offer 379 $68,792 $0 $36 $96 $182 $228 $1,841 Nineth)lowest)offer 348 $64,236 $0 $35 $99 $185 $241 $1,785 Tenth)lowest)offer 319 $60,025 $0 $35 $106 $188 $260 $1,772

7.4 Savings by switching to the least expensive offer from the existing retailer

Figure 43 compares how much customers would save if they switched to the least expensive offer in the market or the cheapest offer from their own retailer. The median of the saving is typically the largest for the three largest retailers. For the largest retailers, the range of savings that their customers might achieve by switching to their best offers is wide. But the median of the range is not large. By contrast, for these same retailers, the range and the median of the saving if their customers switched to lowest offers in the market, is large.

Figure 43. Distribution of saving ($ per customer per year) by switching to cheapest market offer versus lowest existing retailer offer

The dot plot in Figure 44 shows the median annual saving for the least expensive offer in the market compared to the least expensive offer from the customers’ existing retailers. This illustrates clearly the large gap between the two, particularly for the largest retailers.

Figure 44. Median saving ($ per customer per year) by switching to the lowest market offer versus lowest offer from customers’ existing retailers

Analysis of the savings that customers might obtain by switching to the lowest offer from their existing retailer rather than the lowest commonly available offer in the market, suggests that most customers, particularly those supplied by the largest retailers24, would pay considerably less by switching to the lowest offer in the market rather than the lowest commonly available offer from their retailer. This is shown in Figure 45.

24 Alinta and to a lesser extent Click are exceptions but should be treated with caution considering the limited number of their bills in the sample.

Figure 45. Relationship between switching saving and retailers’ market share in actual bill sample

Finally, it may be the case that retailers would be prepared to offer customers lower prices than in their commonly available offers in order to retain those customers if they threatened to switch25. However the data from the sample of bills does not provide evidence that this is common.

25 For the larger retailers that obtain economies of scale in retailing and economies of scope through vertical integration, it is very likely that the marginal cost to supply is far below their commonly available prices. In this case, a profit-maximising retailer can be expected to discount its commonly available prices more heavily in order to retain the customer.

Appendix A: Data and analytical tools

The data used in Part A is contained in MarkIntell (Offer) and the analysis of this data was mostly undertaken in MarkIntell (Insight) with post-processing in Excel in some areas. MarkIntell (Offer) and MarkIntell (Insight) are subscription web-deployed data and analytical tools:

• MarkIntell (Offer) (MIOffer) is a database with all relevant price information, solar feed-in information, GreenPower rates, exit fees and incentives for all publicly available standing and market offers to residential and small business electricity users in Victoria, New South Wales, South East Queensland and South Australia since June 2016 (since April 2016 in Victoria). Until April 2017 all data was sourced by scraping and mining the energy price fact sheets in all licensed retailers’ websites. Since 1 May 2017, MIOffer contains all offers scraped and mined from the two official price comparison websites (which contains offers supplied to them by all licensed retailers). MIOffer is updated daily. MIOffer is updated twice each day.

• MarkIntell (Insight) (MIInsight) uses the data stored in from MIOffer and also includes all relevant network tariffs and matches of these to each retail offer to analyse electricity bills and break them down into their component parts using assumptions on load profiles, consumption levels, wholesale energy and federal environmental certificate prices, customer location and tariff types. Results can be sliced by customer type, distributor, retailer, tariff type, offer type and combinations thereof and displayed using Google Graphics visualisation techniques or exported to CSV for post-processing.

• MarkIntell (Switch) (MISwitch) is electricity and gas price comparison software that captures data from customers’ bills and then uses data from MIOffer and analysis from MIInsight to find the offers that achieve the lowest bills for that customer. It ranks the offers and savings on all available offers based on the consumption profile and existing prices in that customer’s bill. It allows annualisation adjustments of all relevant consumption or solar production parameters to account for seasonal effects. MISwitch covers the contestable

retail electricity and gas markets in Victoria, New South Wales, Queensland and South Australia.

The analysis in Part B was undertaken using SwitchGov, which is an adaption of MISwitch.

Appendix B: Load profile assumptions Value Description Controlled)load)/)dedicated)circuit)/)Annual)consumption)(kWh) 0 NSW)Controlled)load)1)/)dedicated)circuit)/)Annual)consumption)(kWh) 0 NSW)Controlled)load))2/)dedicated)circuit)/)Annual)consumption)(kWh) 0 Solar)export)to)the)grid)/)Annual)consumption)(kWh) 0 Minimum)daily)demand)charge 0 Proportion)of)year)classified)as)summer)in)demand)charge)tariff 0 Peak)demand)(kW/kVA)))summer 0 Peak)demand)(kW/kVA))non/summer 0 Annual)consumption)proportion,)peak)non/summer 0.1 Flexible)Summer)Peak 0.11 Flexible)Non/Summer)Peak 0.13 Flexible)Summer)Off/Peak 0.13 Flexible)Non/Summer)Off/Peak 0.15 Annual)consumption)proportion,)peak)summer 0.2 Flexible)Summer)Shoulder 0.23 Flexible)Non/Summer)Shoulder 0.25 Annual)consumption)proportion,)off/peak)non/summer 0.25 Time)of)use)7day,proportion)annual)consumption,)off/peak 0.26 Annual)consumption)proportion,)off/peak)summer 0.45 Time)of)use)5day,proportion)annual)consumption,)off/peak 0.47 Summer)proportion)in)seasonal)flat)rate 0.5 Non/summer)proportion)in)seasonal)flat)rate 0.5 Time)of)use)5day,)proportion)annual)consumption,)peak 0.53 Time)of)use)7day,)proportion)annual)consumption,)peak 0.74 Proportion)of)annual)gas)demand)in)Off/peak)months 0.8