Energy Price Formation in Wholesale Electricity Markets
The focus of Working Group C5.28 is on how energy price formation is conducted in the various markets around the world. Seventeen countries or regions responded to the survey created by the working group. This paper describes the backgrounds of each of the respondents’ countries or regions, a high-level overview of the market designs as it pertains to price formation, a description of how prices are calculated in each area and then a brief listing of factors that may influence future changes in price formation methodologies in each area.
Members
Convenor
(US)
A. KEECH
Secretary
(US)
N. TACKA
H. AGABUS (EE), M. AL-HAMAD (BA), J. BESSELING (NE), K. BHATTACHARYA (CA), R. CASTRO (US), J. DE LA VIÑA (PH), G. DOORMAN (NO), G. LABUTIN (RU), S. MUKHERJEE (IN), D. PÉREZ OROZCO (CO), R. SACCHI (BR), M. SAGUAN (FR), G. THORPE (AU), N. TRANCHO TEJERINA (ES), P. VALENZUELA (CH), T. YAMASHITA (JA)
Corresponding Members
R. BARROS (BR), R. BRODER HYTOWITZ (US), A. CASTRO SÁNCHEZ (CO), G. DE MIJOLLA (US), A. GIACOMONI (US), A. LAMADRID (US), C. MADINA (ES), I. NDIAYE (US), G. RAMALHO (BR), Y. TAKAMIZAWA (JA)
Responding countries/regions include:
- Australia (NEM)
- Brazil
- Canada (Ontario)
- Chile
- Colombia
- Estonia
- France
- Gulf Cooperation Council (Kuwait, Saudi Arabia, Bahrain, Qatar, United Arab Emirtes, Oman)
- India
- Japan
- Netherlands
- Norway
- Philippines
- Russia
- Spain and Portugal (joint response)
- United States (ERCOT, MISO, NYISO, PJM)
Market Models and Price Formation
The market models of the responding countries tend to fall into three general categories, with variations in the timing of pricing and the spatial granularity of pricing:
- Real-time Locational Marginal Pricing (LMP) models
- Day-ahead zonally priced models
- Uniform pricing models
Importantly, the timing of the market for which each country responded is also different. The survey sought to identify the pricing methodology for the market that is most closely tied to real-time system operations. This is different by each country/region. For example, some countries only have real-time markets while others only have day-ahead markets.
Real-time LMP Models - Russia, Australia, Philippines, United States, Chile
The real-time LMP-style models are generally characterized by real-time energy markets that are closely tied to real-time dispatch. In these models, security constraints within the network itself are modeled in both the dispatch and pricing and as a result these markets have nodal or zonal prices that reflect the cost of energy and the impact of network limitations. Ancillary services are sometimes dispatched and priced simultaneously with energy.
Some things that differ across these models are the structure of supply offers that are used to derive prices. Specifically, US market models that utilize LMP only include a portion of generator costs in the calculation of the price. This can lead to the need to uplift payments to ensure the markets are not confiscatory and further parallel payments such as capacity payments. This can be contrasted with models such as Australia where supply offers represent the “all-in” cost of a resource and therefore there is no need for uplift payments or parallel payments.
Day-ahead zonally Priced Models – Spain, Portugal, Estonia, France, Netherlands, Norway, Brazil, Japan
Day-ahead zonally priced models are the category of markets that are implemented by various countries in the EU. Responses for these market models focus on Day-ahead Market pricing as real-time trading in these markets do not result in price changes. In the EU model, network constraints are modeled and optimized but only between countries/regions rather than within them. Therefore, each country has a uniform price [1] but each country’s price may be different from another.
Prices in these models are derived by bids and offers submitted by market actors that are optimized in a manner that maximizes overall welfare while reflecting the inter-tie constraints between each region.
Because prices in these models do not tie directly to the real-time operation of the market, actions taken by the Transmission System Operator (TSO) to maintain real-time reliability are paid through uplift payments.
Uniform Pricing Models – Colombia, Gulf Cooperation Council (Kuwait, Saudi Arabia, Bahrain, Qatar, United Arab Emirates, Oman), India, Canada (Ontario)
This model contains several respondents whose country or region has one single uniform price. The market-clearing mechanisms in these models differ by region and timing (day-ahead/real time). Because these models do not tie directly to the real-time operation of the market, actions taken by the Transmission System Operator (TSO) to maintain real-time reliability are paid through uplift payments.
Uplift and Parallel Payments
The paper will explore uplift and parallel payments made in the various countries and regions. Uplift and parallel payments will include the various payments made to market actors such as capacity payments, reserve payments, and uplift payments.
Market Performance and Drivers of Change
Finally, the paper will look forward at potential drivers of future change in the price formation processes in each of the areas. There were various causes given in this area. Several common ones include:
- Carbon abatement policies
- Integration of nascent technologies including electric vehicles and storage
- Addition of new grid services into the co-optimized energy market
- Demand-side participation changes
- Changes driven directly by the regulator
Final Thoughts
A primary takeaway from this analysis is that successful markets are designed in various ways and the price formation methodology is dependent on the priorities of the regions. LMP-based markets tend to focus on conveying the current physical state of the system and marginal cost of electricity through their pricing whereas zonal and uniformed priced models tend to have broader pricing regions to facilitate trading between market actors. Within LMP model group, some have nodal prices and some zonal. Zonal designs strike a balance between economic costing and trading and may also be used where there is limited congestion within an electrical region of the market where there are limited difference between nodal prices. Both models can and have been implemented effectively depending on the priorities of the region.
Given that various models have been implemented effectively, it is critical that a region that is looking to start a market, or an established market reviewing its pricing regime, not simply adopt the model used in another region without devoting critical thought to its priorities, the effects on different stakeholders and the physical realities of its system, Areas without congestion or without siginificant congestion may not benefit from the additional complexity of nodal pricing whereas regions with significant congestion may desire nodal pricing to ensure a resource’s incentive to follow dispatch is conveyed through the prices. A further consideration in the pricing regime, not analyzed in detail in this paper, is the existence of a capacity market or other parallel payments. Areas without capacity markets may need to consider incorporating all resource costs into offers and prices whereas those with capacity markets may see benefit from only allowing operational costs into prices. Capacity markets come along with their own set of complexities and so though must be placed on this decision spectrum and balanced with the needs and desires of the region.
While all of the models explored herein have benefits and drawbacks, they all face the potential for future changes due to many factors. The factors motivating change that were most commonly reported in the surveys for this publication were the changing resource mix from fossil resources to renewables, region-specific carbon abatement policies and changing fuel and commodity prices.
- [1] There are some countries (e.g. Sweden, Norway Italy) with different price zones, but the boundaries of those zones are permanent, so, for the purpose of this classification, the statement is still valid.