Chapter 10
Public Policy Issues Associated With Feed-In Tariffs and Net Metering: An Australian Perspective
Darryl Biggar*
Abstract
It has long been known that time-averaged and geographically averaged retail tariffs distort end-customer’s incentives to use and invest in appliances and energy efficiency. This problem is even more acute when end-customers can invest in and use on-site generation and storage. What tariff should be paid for the output of that generation or storage? Should the customer be able to use that output to offset his/her own consumption? Should the customer be able to sell any surplus to neighbors at the local retail rate? This chapter shows that with current retail tariffs, these policies have both pros and cons. Net metering and peer-to-peer trade will remain controversial as long as retail tariffs are not cost reflective.
Keywords
Australian regulations
gross metering
net metering
virtual net metering
group net metering
community shared solar
buy all/sell all
peer-to-peer trade in electricity
1. Introduction
As many chapters in this book have emphasized, the electric power sector is experiencing a fundamental transformation. The rapid uptake of distributed energy resources, energy management solutions, and energy storage has opened up new opportunities for electricity customers, particularly customers connected to distribution networks and on the grid’s edge. Around the world there is an increasing interest in allowing small consumers to produce their own electricity, and not just to sell that electricity back to the grid, but to actively trade that electricity with others, through peer-to-peer trading, open platforms, community-shared solar schemes, and so-called group or virtual net metering.
Many electricity consumers believe they have a right to self-produce electricity if they are able to do so. It is only a small further step to seek to trade that self-produced electricity with neighbors. However, these proposals raise a host of deep and complex issues for policymakers, which are increasingly being raised in Australia,1 and in many other countries around the world.2
As shown in this chapter, the associated policy issues of net versus gross metering, virtual net metering, and peer-to-peer trade in electricity come down to questions of tariff design. Historically the tariffs paid by smaller electricity consumers have not been efficient or cost reflective. The proliferation of on-site generation and storage options places these historic tariffs under considerable strain. The problem is exacerbated when end-customers seek to trade their self-produced or stored electricity with neighbors. A decision will have to be made: either to abandon the historic structure of the network tariffs, or to halt the proliferation of embedded generation and storage and prevent peer-to-peer trade in electricity. The resolution of that tension will determine the course of the electricity industry over the next few decades.
In developed countries, such as Australia, almost all electrical loads (customers) have the right to connect to the grid at their desired location and to purchase electricity at the prevailing retail tariff.3 This chapter starts with the assumption that, subject to meeting technical requirements, embedded generation has a similar legal and physical option to connect to the distribution network and to sell its output at some rate, such as the prevailing wholesale spot price, the local feed-in tariff, or the local retail rate.
This assumption is important. It implies that the underlying public policy issue is not the ability to trade the output of the embedded generation; after all, as long as the overall power system is in balance, every unit sold by the embedded generation must be, in effect, purchased by a consumer somewhere. Rather, the primary issue is the price paid for the output of that embedded generation: both the level and structure of that price, and its level and structure relative to the price paid by load. It is these pricing issues that make the handling of embedded generation and peer-to-peer trading politically tricky.
This chapter takes as given that the overall public policy objective—as set out in Australia’s National Electricity Objective4—is the efficient use of and investment in the electricity sector. This chapter also assumes that all external environmental effects, including the cost of carbon pollution and/or the use of environmental water, are internalized through existing mechanisms, such as taxes, fees, or carbon-pricing mechanisms. This assumption allows us to treat all forms of generation on a level playing field.
This chapter consists of three sections in addition to this introduction. The next section briefly sets out the theoretically efficient tariff design and the implications for net metering and peer-to-peer trade. Section 3 then looks at how differences in the retail tariffs for loads and for embedded generators affect the incentives of end-customers to either combine generation with load (as in net metering) or to separate generation from load (as in gross metering). In practice new embedded generation in Australia is paid a tariff well below retail tariffs, creating strong incentives for net metering. Section 4 looks at the problems that arise as a result and the various options for solving those problems: requiring separate metering for embedded generation, on the one hand, or extending net metering to neighbors, on the other. Section 5 concludes.
2. Efficient tariffs for generation and load in theory
As noted in the introduction, the public policy issues associated with virtual net metering and peer-to-peer trading in electricity come down to questions of tariff design. In particular, these public policy issues depend critically on the design of two tariffs: the tariff paid for the output of the embedded generation, and the tariff paid for electricity consumed by loads.5
In particular, the public policy issues arise from:
• the efficiency (cost reflectivity) of each of these two tariffs; and
• the difference between the tariff for generation and load.
It is useful to be clear at the outset what theory suggests is the optimal tariff. Economic theory is quite clear that first best pricing for generation and load involves tariffs that are fully cost reflective and for which there is no difference in the tariff paid to generation and the tariff paid by load. Efficient tariffs would involve every load being charged (and every generator being paid) the short-run marginal cost of the production and the delivery of a unit of electricity to the location of that customer.
As is well known from wholesale electricity market pricing theory, the marginal cost of production of electricity depends on the marginal cost of the “marginal” generator or load, and varies continuously according to the level of demand relative to the stock of available generation. At off-peak times, when there is plenty of surplus capacity, the marginal cost of producing electricity is low. At peak times, when additional peaking generation must be brought on-line, the marginal cost of generation can be very high.
The marginal cost of delivery of electricity depends on the presence of congestion on the transmission and distribution network. At times when congestion is absent the marginal cost of delivery of electricity is very low, and limited only to the electrical losses on the system. At times and places where congestion is present, the marginal cost of delivery may be very high, leading to substantial price differences in the marginal price of electricity at different locations.
In theory, the marginal cost of production and the marginal cost of delivery can be determined jointly through a market process, which yields a distinct price at each location or node on the network and at each point in time. These prices are known as locational marginal prices6 or nodal prices.
The mechanism to determine such prices is well known, and is now quite routine at the high-voltage level in liberalized electricity markets around the world. This process requires a centralized market operator who accepts bids and offers from market participants and who seeks to maximize the value of trade, subject to the physical constraints of the physical power system. Research is underway in extending these principles to the lower-voltage distribution network level.7
This theory is clear that at any given location on the network, and at any point in time, the price paid for power injected into the network at that location should be equal to the price paid for power withdrawn from the network at that location. In such a market there would be no restrictions on net metering or peer-to-peer trade: any market participant could, in principle, trade electricity with any other market participant, subject to the constraint that all trades would be mediated via the market operator, and all market participants would be paid or would receive their own local locational marginal price. As already noted, these processes are commonplace in the high-voltage or transmission component of liberalized electricity markets around the world.
Locational marginal prices are dynamic and may vary widely from one location to another. Many end-customers will want to be protected from the volatility in the wholesale prices. In the Australian market, this function is performed by retailers. Retailers typically use hedge products to convert volatile wholesale prices into the time-averaged retail prices that end-customers seem to desire. As locational marginal prices are extended down to the distribution network level it will be important to ensure that retailers have the hedging tools they need to be able to convert the volatile wholesale prices into the retail contracts that consumers want, including flat or time-averaged retail prices if that is what consumers desire.8
3. Retail tariffs for generation and load in practice, and their implications
In practice, of course, existing tariffs paid by loads, and existing tariffs paid to embedded generators, depart very significantly from this theoretical ideal.9
3.1. Existing Retail Tariffs for Loads
Retail tariffs paid by loads depart from the marginal cost pricing ideal in two main ways: the presence of substantial fixed costs that are typically recovered through the variable (per kWh) retail charge; and the ubiquitous practice of time averaging and geographic averaging of retail rates.
Historically several, largely fixed, costs have been recovered through variable, energy-based (per kWh) surcharges on tariffs paid by load. There are several such fixed costs:
• the fixed costs of the transmission and distribution networks;
• the costs of running the wholesale market (the cost of the system operator);
• retailer costs; and
• the costs of government schemes to promote renewable generation, such as the Renewable Energy Target.
These costs are summarized in Fig. 10.1 from a report by the public utility regulator in the Australian state of Victoria.
Figure 10.1 Composition of the retail price of electricity (Victoria, Australia).
FiT, Feed-in tariff; RET, Renewable Energy Target; VEET, Victorian Energy Efficiency Target.
FiT, Feed-in tariff; RET, Renewable Energy Target; VEET, Victorian Energy Efficiency Target.
In addition, retail tariffs for load in Australia have been highly time averaged and geographically averaged. Furthermore (although perhaps less significantly) retail tariffs typically include sales taxes or value-added taxes, which are not paid on self-produced goods and services.
When it comes to tariffs for embedded generation in Australia, a few such generators sell their output in the existing wholesale spot market. However, most small-scale embedded generators in Australia are paid a simple time-averaged feed-in tariff. As set out in Fig. 10.2, although some of these feed-in tariffs started out much higher than the retail tariff, over the past few years the feed-in tariff offered to new solar PV customers has declined rapidly10 and now typically reflects the avoided cost of wholesale generation (on the order of 6–8 c/kWh).11
Figure 10.2 Evolution of feed-in tariffs 2008–2015 Australian states.
ACT, Australian Capital Territory; NSW, New South Wales; QLD, Queensland; SA, Southern Australia; Tas, Tasmania; Vic, Victoria; WA, Western Australia.
ACT, Australian Capital Territory; NSW, New South Wales; QLD, Queensland; SA, Southern Australia; Tas, Tasmania; Vic, Victoria; WA, Western Australia.
3.2. Implications of Differences in the Tariff for Generation and the Tariff for Load
Differences in the price paid for the output of embedded generation and the price paid for electricity consumption at the same location and the same point in time give rise to strong and often highly undesirable incentives. The precise incentives that arise depend on whether the generation price is above or below the load price. The simple mathematics of these results are set out in Box 10.1.
Box 10.1 Impact of Differences in Tariffs on Incentives
Let’s suppose that, at a given site, the load at a point in time is QL and the generation on the same site is QG. The price paid by the load is assumed to be PL and the price paid for the output of the generation on the same site is PG. The total amount paid by this customer is therefore:
Let’s assume that the system operator cannot monitor the output of the embedded generation and the on-site load directly. Instead, the system operator can only easily monitor the net consumption at the site, which we will label N = QL – QG. The end-customer is assumed to be able to manipulate the apparent on-site generation and load subject to the constraints that the net consumption is given by N and the on-site generation and load are both required to be positive. The total amount paid by the customer can be written in two other equivalent ways:
In the case where generation price is equal to the load price PL = PG, the total amount paid is just the common price multiplied by the net consumption PGN = PLN. In this case the customer cannot manipulate the apparent generation or consumption (while holding the net consumption fixed) in such a way as to increase or reduce his/her payment.
However, when the generation price is different to the load price the customer may be able to manipulate his/her payment by altering the apparent local generation and load. There are two cases to deal with:
• When the generation price is less than the load price (PL > PG), the previous equations above show that the customer would like to make both the volume of generation QG and the volume of load QL as low as possible (while holding the net consumption N constant). This can be achieved by using on-site generation to offset on-site load. If the customer is a net importer of power (N > 0), the best the customer can do is to make the on-site generation appear to be zero QG = 0, in which case the customer pays the load price PL for any increase or decrease in the net consumption of power (from the second equation). On the other hand, if the customer is a net exporter of power (N < 0), the best the customer can do is to make the on-site load appear to be zero QL = 0 in which case the customer pays the lower generation price PG for any increase or decrease in the net export of power (from the first equation).
• On the other hand, when the generation price is above the load price (PL < PG), the customer would like to make both the volume of generation QG and the volume of load QL as high as possible (while holding the net consumption N constant). This could perhaps be achieved by separate metering of generation and load (i.e., gross metering) and by diverting load through the generation meter (in the manner suggested in the text). In principle there is no upper limit on the revenue the customer can receive by arbitraging in this manner.
Let’s consider first the typical case in which the price paid for the output of embedded generation is less than the price paid by load. For example, a retail load in Australia may pay a price of around 27 c/kWh. However, a generator on the same site may only be paid the avoided wholesale cost (around 6 c/kWh).
In this case the owner of the embedded generator has a strong incentive to sell directly to load in the vicinity, rather than sell its output to the wholesale market at the feed-in tariff (Box 10.1). This can be achieved by:
• using the output of the local generation to offset local load (which is referred to below as a form of net metering),
• constructing a physical duplicate network to link up the generation with neighboring local loads (even where such duplication is inefficient), or
• establishing a separate local embedded network combining generation and load.
To summarize, when the generation price is lower than the load price, it can be said that the generation has a strong incentive to combine with the load as much as possible. These actions can undermine legitimate public policy objectives, as explained further below.
Now let’s consider the opposite case in which the price paid for the output of embedded generation is greater than the price paid by load at a point in time. This might be the case for customers on a “premium” feed-in tariff,12 or for generators that are paid the time-varying wholesale spot price, at times of high wholesale prices.
In this case there are strong incentives to separate generation from load. This may involve:
• Gross metering, that is, the separate metering of the output of the generation from the load (as set out below).
• Breaking up a microgrid or embedded network into its constituent parts, even where it is inefficient to do so.
• Diverting load so as to appear as though it is generation. One possible way this could be achieved is for the load to be used to charge batteries (paying the lower load price), and for the output of the batteries on discharge to be used to supplement the output of the embedded generation (receiving the higher generation price). This will be referred to as “load masquerading as generation.” In principle, there is no upper limit on the extent to which the customer could increase revenue by arbitraging in this way.
These results are summarized in Table 10.1.
Table 10.1
Differences in Prices for Generation and Load Affects Incentives of End-Customers
| Case | Incentive |
| Generation price below load price | Generation seeks to combine with load, using net metering, colocation, or the creation of a duplicate physical network (such as an embedded network or microgrid) |
| Generation price above load price | Generation seeks to separate from load using gross metering, separate location, or diversion of load to masquerade as increased generation |
4. Current problems and possible future directions
4.1. The Status Quo and its Problems
In Australia the price paid for the output of embedded generation (the feed-in tariff) is typically much lower than the price paid by retail load.
In addition, for reasons discussed further below, in practice it is very difficult for an electricity network or retailer to identify and separately meter the output of embedded generation against the will of the site owner. Typically the broader network will only be able to meter the net consumption or the net exports of each site.
The status quo in Australia therefore involves the following three elements:
1. Retail tariffs for load that are both inflated and highly time averaged.
2. A price for net exports that is time-averaged and much lower than the price paid for net imports.13
3. Difficulty (or impossibility) of separately identifying and metering embedded generation.
This combination of elements gives rise to several, material public policy issues relating to the use of and investment in embedded generation:
• There are inefficient incentives for both the use of and investment in embedded generation. End-customers do not have the correct incentive to use their embedded generation at times when it is most highly valued, and they do not have an incentive to choose the right type of generation (specifically, generation that can best respond to wholesale market conditions). In particular, the private incentive to install solar PV may be much larger than the social benefit of installing solar PV generation.14
• There are inefficient incentives regarding the sizing of embedded generation. End-customers have an incentive to install embedded generation sufficient to offset only their own load (unless they can combine with other customers to form an embedded network, as noted in the next bullet point). This may result in inefficient sizing or scaling of embedded generation.
• There are inefficient incentives regarding the establishment of embedded networks; end-customers have an incentive to create separate local networks combining load and generation, even if doing so involves duplication of existing network infrastructure.
• There are inefficient incentives regarding investment in storage. As the price paid for net exports is lower than the price paid for net imports, end-customers with sufficient embedded generation to enable net exports at times have an overly strong incentive to install storage, so as to arbitrage between times of net exports and times of net imports, even if there is little or no social benefit arising from that storage.15
• The increasing penetration of embedded generation poses a threat to the revenue stream of network operators. As long as fixed costs are recovered through inflated variable (usage) charges for electricity, increased output of embedded generation reduces the revenue received by the network business without a corresponding reduction in its costs. This may result in higher prices for all customers, including customers without embedded generation, further stimulating investment in energy efficiency and embedded generation, and increasing the risk of a death spiral.16
There are also potential arguments that the status quo is inequitable or unfair because not all end-customers have equal access to embedded generation. Some customers, such as those with ample roof space, can enjoy the benefits of reduced electricity charges while others, such as those in apartment buildings, cannot.
4.2. Gross Metering, if it Were Feasible, Could Partially Solve These Problems
What can be done to solve these problems? We noted above that the problems stem from three underlying elements: inefficient tariffs for load, tariffs for generation less than the tariff for load, and the difficulty in separately metering the output of on-site generation from the output of on-site load.
One possible solution, of course, would be to insist—despite the difficulty—that any on-site generation must be separately metered from an on-site load. This is known as gross metering.17 The definitions of gross metering and net metering are set out in Table 10.2. This possibility is illustrated as “solution 3” in Fig. 10.3.
Table 10.2
Definitions of Gross and Net Metering
| Gross metering | The on-site generation and the on-site load are separately identified and metered, from the first unit of output or consumption in the same way as would arise if the generation and load were located on separate sites |
| Net metering | The on-site generation is required to be offset against on-site load at the same point in time. Net imports to the site are charged at the standard retail tariff for that point in time. If there are net exports from the site, the amount paid depends on the form of net metering which is discussed further below |
Figure 10.3 Summary of the key policy issues.
To many end-customers a requirement that the output of embedded generation be separately metered and paid a lower price seems wrong. Many people have a strong sense that, where it is feasible, they should be allowed to produce on-site to meet their own needs, whether in electricity or in other services. When the Queensland Government considered adopting gross metering for solar PV installations, the response from some solar PV owners was one of disbelief. The Policy Director of the Clean Energy Council observed:
What the Queensland Competition Authority has proposed is the equivalent of telling people they can’t just use the lemons growing on the lemon tree in their backyard – they have to sell the produce to a wholesaler for next to nothing, and then buy the lemons back at a premium from the supermarket.18
In fact, as emphasized above, there are circumstances where a government might consider telling customers they cannot directly consume the lemons they grow in their backyard. As long as the retail tariff for load is artificially inflated and time averaged, a requirement for separate metering for on-site generation has several potential benefits in partially addressing the problems identified above. Specifically, gross metering potentially:
• allows for improved price signals to embedded generation (while preserving the existing flat retail tariff for loads), potentially increasing the efficiency of both the use of and investment in embedded generation;
• eliminates the incentive to artificially size on-site generation, allowing for more efficient larger-scale generating installations where it is feasible;
• eliminates the incentives to create artificial embedded networks and to inefficiently bypass the existing distribution networks;
• eliminates the incentive to install inefficient storage facilities to arbitrage between times of net exports and times of net imports; and
• preserves the contribution to fixed charges embedded in the retail tariff.
On the other hand, as noted earlier, gross metering may simply be infeasible or impossible to enforce. There is no sensible distinction between an increase in on-site generation and a reduction in on-site load. Any attempt at defining on-site generation for the purposes of gross metering will result in artificial boundary issues, which are difficult to police.
For example, what constitutes a “generator” for which separate metering is required? Must a generator always export power? If gross metering is imposed, manufacturers may seek to install small-scale generation or storage inside consumption devices. Consider for example a pool pump or an air-conditioning unit that normally draws from the grid, but which has the capability to switch to an internal battery at peak times. Should such devices be labeled as generators or loads? The distinction is entirely artificial.
Even more importantly, without costly detailed on-site inspection, a network or retailer can not know whether or not a customer has installed an on-site generator.19 Historically solar PV installations have tended to be more visible than other forms of on-site generation, but even these could be hidden in the future20 and, in any case, it would make no sense to target one form of embedded generation over others. Even if an on-site generator could be identified, it is likely to be extremely difficult to inspect and verify the way the generator is physically wired, to ensure that the full output of the generator passes through the meter, rather than being used to offset on-site load.
There is also the difficult question of how to handle storage, which is both a consumer and a producer of electrical energy. Should battery storage be charged the load rate when it is charging, and the generation rate when it is discharging? Would this require two separate meters? How would this be enforced?
Although gross metering has some economic merit over the status quo, it is not clear that it is feasible in practice.
4.3. Extended Forms of Net Metering Could Also Partially Solve These Problems, but Each has Problems of its Own
Rather than a separate metering of generation and load, perhaps the solution is to allow further integration of generation and load, allowing generation not just to offset on-site load, but also loads located on neighboring sites. This possibility is illustrated as “solution 2” in Fig. 10.3.
At one level, allowing local generation to offset neighboring loads does not seem unreasonable. After all, this generation is already physically connected to the neighboring loads through the local distribution network. These arguments seem particularly compelling where the end-customer owns the neighboring buildings,21 or where the neighboring buildings don’t have the capability to install embedded generation of their own.
The arguments may also seem reasonable where allowing such trade allows for the efficient sizing of an embedded generation facility. If a customer is allowed to install solar panels on their own roof to offset their own load, should they not be allowed to receive the same (or larger) benefit by paying for a share of the larger, more cost-effective, community-scale facility at the end of the street?
These questions amount to asking the question: with whom should an embedded generator be allowed to trade “directly” (without going through the market operator). Should an embedded generator be only allowed to trade directly with on-site load? With designated neighboring loads? With all neighboring loads? Or, with all loads? These questions can be thought of as different forms of net metering. There are at least four different net metering options, summarized in Table 10.3.22
Table 10.3
Definitions Net Metering Used in This Chapter
| On-site net metering | The on-site generation is only allowed to be offset against on-site load, as long as there are no net exports from the site. Net imports to the site are charged at the standard retail tariff. If there are net exports from the site, the amount of the net export is paid the feed-in tariff. This form of net metering is standard in Australia |
| Virtual net metering | On-site generation is allowed to be offset against on-site load and other designated loads in a designated neighborhood at the same point in time, such as the participants in a community-shared solar scheme. Other net exports are paid the feed-in tariff |
| Local or community net metering | On-site generation is allowed to be offset against all load in a designated site, campus or neighborhood at the same point in time, such as a shopping center or housing subdivision |
| Global net metering | On-site generation is allowed to be offset against all loads at the same point in time |
In the United States the most common form of net metering allows a customer to “bank” credits for exported generation and use those credits to offset the customer’s own consumption at later points in time.23 Another version of net metering sees the customer paid “the average retail utility energy rate” for these export credits. Using this terminology above, these are both forms of global net metering, with slight variations in the tariff paid for net exports, although with limits on the total amount of exports that can earn the retail tariff rate.24
Some states, such as California, allow the owner of a multimeter property to use credits for net exports from one customer to offset the energy bills of other tenants in the same building. The California Public Utility Commission refers to this as virtual net metering.25
These extended forms of net metering could partially solve some of the problems identified above. Specifically, extended forms of net metering could potentially:
• Reduce the incentive for inefficient network bypass, by eliminating the incentive to build a physical network between generation and load.
• Reduce the incentive to create artificial embedded networks, instead allowing the creation of “virtual” embedded networks.
• Reduce the incentive to inefficiently scale the embedded generation. If the end-customer is able to find sufficient load to offset the local generation, the local generation can be efficiently sized.
• Reduce the incentive to install inefficient storage. If the end-customer is able to find enough load to offset the local generation, there may be little or no output that is paid the net export price, thereby reducing the incentive to install inefficient amounts of storage.
In addition, extended forms of net metering could be argued to reduce the apparent inequity or unfairness in access to embedded generation opportunities. Extended forms of net metering allow all customers (or at least all customers in the proximity of an embedded generation facility) to share in some of the benefits of that facility, whether or not they are able to install a generating facility of their own.
However, allowing for extended forms of net metering exacerbates the problems identified above. Specifically, extending the scope of net metering will:
• not improve (and potentially worsen) the price signals on the embedded generation, which leads to inefficient usage and investment decisions in embedded generation26; and
• exacerbate the problem of erosion of the revenue streams of network operators by reducing the contribution to various fixed costs, including the fixed costs of network provision.27
These problems have led in the United States and Canada to the placing of limits on the volume of embedded generation that can qualify for net metering, either on an individual site (kW) or collectively (by the number of subscribers or the total volume of embedded generation).28
There is another flaw with extending the scope of net metering, which is potentially even more serious. As noted at the outset, the efficient price for both generation and load is a price that reflects the short-run marginal cost of producing electricity and delivering that electricity to the location of the customer. As long as the generation and load are located at the same physical site, the price paid by load and the price paid for generation should be the same. However, as long as the generation and load are physically separated there is at least a possibility that the marginal cost of delivering electricity to those two sites is different.
The marginal cost of delivering electricity to location A may differ from the marginal cost of delivering electricity to location B due to either losses or congestion. The further apart the sites are, the greater the likelihood that the electrical losses will differ from one site to the other, and the greater the likelihood that, from time to time, congestion will arise on the network on some path between A and B. Such congestion prevents the carrying out of a simple trade between A and B.
Extending net metering from on-site net metering to neighboring locations effectively sets the generation price at location A equal to the load price at neighboring location B. This is only efficient where there are no losses or network congestion between those two locations. This is possibly a reasonable assumption for small subnetworks, such as within a building, shopping center, campus, or subdivision. It becomes less reasonable the greater the geographic separation between the locations, and therefore the greater the scope for congestion to affect the locations differentially.
The failure of these extended forms of net metering to reflect congestion on the local network is a major drawback. It raises the possibility that local trading of electricity will be incompatible with the physical capability of the local power system. This will at least be inefficient and may, at worst, force the intervention of the system operator to maintain system security.29
To summarize, extended forms of net metering are essentially a local form of electricity trading that ignores the physical limits of the local distribution network. Such forms of peer-to-peer trading can only ever be possible over a relatively small geographic area. The larger the area, the greater the likelihood that the resulting trades will be incompatible with the local network, forcing the system operator to intervene. If there is to be large-scale, sustained peer-to-peer trading in electricity, it will only be possible through a mechanism that integrates that trading with the physical limits of the power system. Such mechanisms already exist at the high-voltage (transmission) end of the market in liberalized electricity markets around the world. If peer-to-peer trading in electricity is to be sustained, it is essential that such mechanisms be extended to include smaller generators and loads.
5. Conclusions
Around the world there has been substantial interest in facilitating the introduction of small-scale embedded generation into local distribution networks. This process has been eagerly promoted by advocates who see small-scale embedded generation as environmentally sound and as giving a degree of control, autonomy, and choice to the small consumer. It is a small step from these developments to envisage local peer-to-peer trading in electricity, particularly over private embedded networks, community schemes, and microgrids. These schemes are said to further enhance environmental outcomes, and to promote community cohesion.
However, historically retail tariffs have been both high (inflated with a contribution to fixed costs) and time averaged. In this world, end-customers have distorted incentives to invest in on-site generation; the incentive is both too strong, doesn’t provide the right usage signals, and doesn’t provide the right incentives to choose the right type of on-site generation. Furthermore, investment in embedded generation threatens to undermine the revenue stream of distribution networks, by reducing their contribution to fixed costs. These effects have been observed around the world.
Under the status quo in Australia, end-customers can use the output of their embedded generation to offset their own on-site load, but not load on other sites. In addition, the tariff paid for net exports is almost always much lower than the tariff paid by net imports. As a result there are strong incentives to limit the size of the on-site generation to match the on-site load, even if that means foregoing economies of scale. In addition, where end-customers cannot control the output of the embedded generation (as with solar PV), the customer has a strong incentive to invest in assets to store the electricity produced at times of net exports to reduce consumption at times of net imports, even if those storage assets have no overall social value. Furthermore, end-customers have an incentive to physically connect embedded generation to local loads, in effect setting up a duplicate local distribution network, or to set up a private embedded network, even where the use of the existing public network is more efficient.
Faced with these problems, policymakers may be tempted to move in one of two directions: The first direction is to insist on gross metering of embedded generation. This would, in principle, allow for a much more efficient tariff for the embedded generation, while preserving the contribution to fixed costs embedded in the current load tariffs. However, gross metering is difficult to enforce in both theory and practice. There is no meaningful distinction between an increase in on-site generation and a reduction in on-site load. To our knowledge gross metering has not been imposed as a regulatory requirement (although it has been allowed in situations where the generation tariff is substantially above the retail tariff).
An alternative direction for policymakers is to allow extensions to the principle of net metering, essentially allowing the embedded generation to trade directly with other loads in the neighborhood. This has some apparent benefits: it allows exploitation of economies of scale, reduces incentives for network bypass, and reduces the incentive to set up separate private embedded networks. However, in a classic application of the theory of the second best, it is not possible to say that extending net metering will be desirable overall, as extending net metering makes the problem of bad price signals for embedded generation worse and worsens the erosion of revenue from the contribution to fixed costs.
The extended forms of net metering discussed in this chapter offer the potential for peer-to-peer or direct trading of electricity between end-customers. However, these extended forms of net metering ignore the potential for local network congestion. The theory is quite clear that direct trade between two parties is only possible under quite extreme assumptions, including the absence of local network congestion. Otherwise, all trade must be mediated through the system operator. It is conceivable that some subnetworks will be designed so as to not feature any internal network congestion. However in the broader network we do not foresee the possibility of unrestricted direct peer-to-peer trade. Attempts to introduce blockchain-based peer-to-peer trading in electricity networks30 will not become mainstream, unless they can integrate the physical limits of distribution networks.
Without further tariff reform there will be an on-going tension between embedded generation on the one hand, and public policy objectives of efficiency and fairness on the other. If net metering continues, or is expanded, embedded generation will likely proliferate, but the objectives of efficiency and revenue stability will be undermined. On the other hand, if net metering is prevented (perhaps insisting on gross metering), it will be possible to improve efficiency in the use of embedded generation, but investment in embedded generation is likely to be restricted by regulatory controls. These public policy trade-offs can only be resolved through a move to more cost-reflective tariffs, such as locational marginal pricing. There are some encouraging early signs of the need to reflect congestion in distribution network pricing in Australia,31 but so far few concrete steps have been taken. The resolution of this tension will set the direction of the electricity industry for the next several decades.
Acknowledgments
Darryl Biggar is the Special Economic Advisor (Regulatory) for the ACCC and the AER. Joe Dimasi is the Senior Commissioner at the Independent Competition and Regulatory Commission, ACT and the Chair, Tasmanian Economic Regulator. The views expressed in this chapter are those of the authors and do not reflect the views of the ACCC, the AER, the ICRC, or OTTER.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.