How best to describe the NEM's pricing model?

Locational price formation in the NEM - nodal, zonal or a bit of both?

Transmission investment along with access reform and locational signals remain front and centre of a number of regulatory reviews and rule changes, bringing up many 'old chestnuts' (MLF, ALF) and a few new ones (CRM, priority access). But before we decide how to move forward, we must know clearly what in lays in before us, in the form of the NEM's current pricing model.

The NEM is often described as a zonal price model - but that does not do justice to some of the nuances of its original design. The devil is very much in the details, and on examining at a more granular level, electricity prices in the NEM are in actuality a form of approximate or quasi- nodal price. This is important as we consider the slew of potential reforms to prices, access and ancillary service markets going forward.

Before that some important theory:

At the core of it is the concept of marginal prices - in electricity markets, the theoretically optimal price signal is the nodal or locational marginal price (see seminal works by Fred Schweppe, Richard Tabors, Michael Caramanis, Paul Joskow, Richard Schmalensee, Hugh Outhred and many others). This is the fundamental justification for the development of the spot price of electricity and the market itself.

If marginal prices are itself your concern, then stop here, and read https://www.dhirubhai.net/pulse/electricity-prices-whats-problem-farhad-billimoria/?trackingId=412wFF9gSwSnkx4oMrZjmg%3D%3D

Nodal or locational marginal electricity prices can be broken down into 3 components:

1. the energy price
2. the price of congestion
3. the price of losses

(each of these is interrelated, and depends upon the choice of reference node.)

The following articles and explainer documents are recommended reading as to how prices are actually formed in the NEM:

From Ben Skinner :

https://www.energycouncil.com.au/analysis/marginal-loss-factors-will-someone-please-repeal-the-laws-of-physics/

From AEMO: https://aemo.com.au/-/media/files/electricity/nem/security_and_reliability/loss_factors_and_regional_boundaries/2016/treatment_of_loss_factors_in_the_nem.pdf

Is the NEM a zonal or nodal model?

I found AEMO's phrasing of prices in the NEM (in the 2016 document above) as "an approximate version of nodal pricing" as quite an apt description of the intent. The NEM adopts a number of approximation and simplifications but it appears the intent was to replicate a nodal price as was most practical at the time. Hugh Outhred and John Kaye describe some of the original justifications for the approximate model in:

H. R. Outhred and R. J. Kaye, “Incorporating Network Effects in a Competitive Electricity Industry: An Australian Perspective”, Chapter 9 in M Einhorn and R Siddiqi (eds), Electricity Transmission Pricing and Technology, Kluwer Academic Publishers, 1996, pp 207-228.

Breaking this down further into its components we find that there is a slightly different treatment in pricing between each of the sub-components. Energy and congestion are priced at a zonal basis to the regional reference node (RRN), while losses are sought to be priced to the node (at the network connection point). (NB: a further wrinkle, due to the hub-and-spoke network model in the NEMDE .. the dispatch engine... nodal losses are estimated through the MLF projection -- see below)

It is also important to recognise that the way in which we approximate nodal prices over time also varies based on the individual sub-components. While we price energy and intra-regional congestion dynamically in real time, different treatment is applied to losses.

Inter-regional lossses (between different RRNs) are dynamic in real time, but intra-regional losses (between the node and RRN) are static and approximated via a fixed annual Marginal Loss Factor (MLF) at the transmission level (and a fixed annual Average Distribution Loss Factor (DLF) for distribution networks).

In actuality both the MLF and DLF are averaged over dispatch intervals in the projection year, which makes them an Averaged Marginal Loss Factor and an Averaged Average Loss Factor... confused yet?

And so we find while the intent was 'approximately nodal' the outturn is a combination of nodal and zonal, dynamic and static components.

The underlying factors that drove this design, as outlined by Outhred and Kaye (1996), include computational tractability, modelling of system security, large-participant problems, limitations to nodal pricing in distribution networks, and tradeoffs between liquidity and precision in the implementation of forward markets. Some are artifacts of the time of liberalisation (the early-to-mid 1990s), while others may continue to remain relevant.

Implicit or explicit hedging?

It is also important to note that by virtue of this design, certain price components attract implicit hedges, while others are unhedged.

Next steps and reform:

In considering any reform to price formation and access in the NEM, we should question which of original artifacts of the NEM's price formation design remain relevant and which do not. We also need to consider the broader strategic directionality of market systems - given the rapidity, flexibility, controllability of system resources, including on the load side.

It is also worth questioning the ongoing rationale for the apparently different implied hedge treatment of different price components - as between losses and congestion. This would only make sense if there are different ways in which participants can manage risks associated with different components - particularly losses and congestion. Alternatively, if losses and with congestion are more relevantly considered as sub-components of a broader financial risk, perhaps a more comprehensive and all-encompassing hedging arrangement is more suited to manage the associated volatility.