Network pricing is not driven by economics, but by policy
For existing customers (connection points in network terms) on a grid with unconstrained substation capacity, that is the vast majority of the electricity grid in the NEM right now, the variable cost to the network of distributing electricity is close enough to zero. The vast majority of the cost was determined at the time the network was built and that cost was mainly driven by the geographic spacing of customers.
Because the variable cost is zero, how networks recover costs is essentially a policy decision. From the network economist’s point of view charging customers on the basis of consumption is no better or worse than charging them based on maximum monthly demand. Neither has much to do with network costs. Governments happily tolerate geographic cross subsidies.
Therefore the policy decision can be made on the basis of what behavior the policy wants to incentivise or disincentivise. Unfortunately this, too, is a can of worms. It’s not just consumer behaviour but also the retailer, generator and network behaviours that need to be considered.
Should policy be designed to encourage customers to be energy efficient? If so, price on the basis of consumption. This has the side effect of being progressive.
Should policy be designed to incentivise customers to buy household/business storage and to minimise the need for grid augmentation? If so, charge customers on the basis of their contribution to maximum monthly demand.
Should policy be designed to encourage micro-grids and sharing excess rooftop PV with your neighbour? If so, charge customers a fixed monthly access fee but don’t charge for consumption or demand. If the fee is too high you incentivise them go off grid. This is the internet or telco model. It’s a regressive model, though, in that less well off customers have to pay the same price as more well off customers.
Should policy be designed to reflect the fact that it’s cheaper to build networks in the city where customers are close together than in rural areas? If so, charge customers based on the actual capital costs of connecting them. This is one area where network costs do matter, but it’s the one that government is most reluctant to touch.
In the end, policies need to facilitate the new electricity economy. Peer-to-peer trading rates should be negotiated, but clearly shouldn’t reflect the fully allocated sunk capital costs.
PV probably contributes more to minimising true network peaks, as opposed to household network peaks, than is sometimes acknowledged.
Distributed storage and PV make the grid more reliable and reduce the total amount of network and transmission hardware that is required, but they may increase the amount of intelligence and software investment needed.
Networks as regulated monopolies is an increasingly outdated idea that poorly aligns outcomes and responsibilities and discourages new ideas. We need a new but not well understood way to make networks more competitive and even more responsive to the way the industry is evolving.
A simple start might be to provide a few more choices to consumers. We are starting to see this with opt-in schemes, but maybe we need a few more – keeping them simple. Pay a demand charge, or a consumption charge, or a fixed charge – your choice.
Network costs are sunk capital costs driven by spacing between connections
Virtually the entire revenue base of a network is in place to recover fixed costs. Most of those costs are capital costs. Those are essentially the costs of owning and maintaining the existing network and the cost of expanding the network for increased numbers of customers. About 37 per cent of revenue is recovery of operating costs, but they are all more or less fixed in the short run.
Figure 1: United energy, FY17 regulated revenue. Source: AER
As Mervyn Davies states, the logical way to recover these costs is buy either a one-off cost at the time the house is connected (and then reflected in the subsequent house or business sale price) or by a fixed annual charge. For networks that are capacity constrained there is also a need to recover a capacity expansion or augmentation cost, but we ignore that in this article.
Although there has been a big focus on both time-of-use and or maximum demand-based tariffs they are both essentially irrelevant to the network’s marginal costs.
Electricity distribution network cost drivers
This section of the note draws very heavily on presentations made to a “cost reflective tariff seminar” organised earlier this year by the APVI. In particular we have used Mervyn Davies presentation on network costs.
Figure 3: Network components: Source Mervyn Davies
The basic message is that for urban networks more than 50 per cent of costs are driven by location, and for rural networks locational costs are over 80 per cent. All of those are sunk costs and economic theory tells us that sunk costs are irrelevant to incremental decisions. It’s the spacing between customers that drives most of the poles and wires cost regardless of demand. And those costs are sunk at the time the infrastructure is installed. Since for any grid you are charged the same price regardless of location, there are embedded cross subsidies. These cross subsidies won’t be removed any time soon but their existence actually is a barrier to storage, micro grids and distributed generation because they hide the true cost of the grid. At the same time customers overpay in the profitable areas, reducing the possibilities for peer-to-peer trading.
Figure 4 Example of uneconomic portions of grid: Source Western Power
The next point to note is that for a “typical” network – and we define that to be United Energy – wires and poles and street level substations make up about 70 per cent of regulated asset value. Those assets are also getting 97 per cent of the “growth” capex, which we define as the excess of investment over straight line depreciation. That growth capex essentially represents the investment for new housing and business developments in this franchise area. One caveat is that the RIN data underlying this table only goes down to the 33 Kv level. Much of the street level infrastructure is at 11 kv and 22 kv.
Figure 5: Rab by class of assets. Source: AER
One way to charge for this service, is as a once-off connection cost or an annuity (monthly charge). As a matter of fact, in the UK they have actually tried to to make the provision house level services contestable. Developers can request a tender for connection and build it into the house price. Just like the internet once you have paid your monthly fee you can use the service as much as you like with no incremental costs. However we will come back to the policy question of whether this is a “good thing” later.
Expected collective demand drives some costs
For that part of costs which are driven it’s the expected collective demand rather than the actual individual demand that is relevant.
Figure 6: Relationship of individual houses to collective household max demand. Source: Passey & Haghdadi 2017 APVI
Residential load is only part of the story anyway
However the bigger point here is that the overall network peak has very little to do with the demand peaks of individual houses. Network peaks tend to be driven by industrial, commercial and household loads all contributing. These peaks are generally in the afternoon, often before household demand peaks. For instance, in the past four years, 45 per cent of the top 100 half-hourly demand peaks in NSW were at 4:00 pm or earlier.
Figure 7: Peak demand driven as much by business and industry as households. Source, NEM Review, ITK
At the moment, consensus thinking about networks is very conventional. Networks are monopolies that should be regulated and ring fenced from competitive pressures. Networks should price their services to “customers” on the basis of SRMC or LRMC, although as we have seen in practice, this can’t or doesn’t happen.
In the end, network pricing is decided by the regulator in what it perceives to be the best interests of consumers.
Network and transmission costs broadly make up 50 per cent of the cost a household pays for electricity and a significant part of the price paid by business and industry.
Networks, though, don’t really have any customers, as such. Networks bill retailers, not the end users, but even the retailers are not really customers. Retailers have very few rights or powers in dealing with networks. End users even less.
The regulatory process is messy and inefficient with information asymmetry (networks have more information than the regulator). The regulator is incentivised to minimise cost and this tends to disincentivise innovation.
In ITK’s view this entire system as increasingly dysfunctional. At a minimum it acts as a barrier to innovation and progress. In the worst case it means that electricity is produced and distributed inefficiently and at a higher than necessary cost.
Thinking differently, how could networks be made more competitive?
The suggestions below are more in the nature of “whiteboard brainstorming” rather than carefully thought out or still less fully developed proposals.
- Network licenses could be made for a set time instead of into perpetuity. The network license would be bid for on the basis of price offered for the service. At the moment when a Government privatizes a network it seeks the highest price for the franchise. It could though sell the network on the basis of the lowest revenue cap offered. This would be analogous to a reverse auction or tender for a renewable project. The issue in this model becomes service quality and the lack of incentive to spend capital.
- Retailers could be allowed to buy blocks of network capacity and then price that to endusers as they please. This would not be a monopoly right but just a slice of network revenue. The retailer could then recover that network revenue from its customers in anyway it pleases.
- New connection investment could be made competitive with the resulting assets owned by point of connection rather than by the network. Under this model the property developer invites bids to supply the wires and poles from the street, or street level substation, to the connection point. The cost of connection is built into the propery price and thereafter usage is “free” to the property owner at least up to some capacity limit. Broadband internet connections work in this way.