|
Article Excerpt
In this article, we use half hourly spot electricity prices and load data for the National Electricity Market (NEM) of Australia for the period from December 1998 to August 2007 to test for randomly modulated periodicity. In doing so, we apply signal coherence spectral analysis to the time series of half hourly spot prices and megawatt-hours (MWh) load demand from 7/12/1998 to 31/08/2007 using the FORTRAN 95 program developed by Hinich (2000). We detect relatively steady weekly and daily cycles in load demand but relatively more unstable cycles in prices.
1. INTRODUCTION
A crucial feature of price formation in electricity spot markets is the instantaneous nature of the product sold. The physical laws that determine the delivery of electricity across a transmission grid require a synchronization and balancing of the input of power at generating points and output of power at demand points together with some allowance for transmission loss associated with electrical resistance and the heating up of conductors. Across the grid, production and consumption decisions must be perfectly synchronized, without any capability for storage, otherwise the quality of supply can be severely compromised. Moreover, while electricity generation and transmission may be viewed as yielding a commodity, its ultimate consumption at the retail end is a service. Thus, the task of either the grid operator or the short-term market mechanism is to continuously monitor the demand process and allocate generating capacity, in line with fluctuations in demand (Bunn 2004, 2, Hinich, Czamanski, Dormaar, and Serletis 2007).
Recently, researchers have applied innovative methods in modelling spot wholesale electricity prices and loads. See, for example, Zhang and Dong (2001), Higgs and Worthington (2003), Deng and Jiang (2004), Leon and Rubia (2004), Serletis and Andreadis (2004), Higgs and Worthington (2005), Lu, Dong and Li (2005), and Worthington, Kay-Spratley, and Higgs (2005). Our contribution here is to offer forecasters a better understanding of the periodicity of prices and load in this market through the use of the Randomly Modulated Periodicity (RMP) model, recently proposed by Hinich (2000), Hinich and Wild (2001, 2005) and applied to the Alberta Electricity market in Hinich, Czamanski, Dormaar, and Serletis (2007). We use this parametric statistical model to study the Australian National Electricity Market (NEM) wholesale spot market. We examine half hourly spot electricity prices (defined in terms of megawatt-hours (MWh)) and MWh load (demand) over the period from 7/12/1998 to 31/08/2007.
Our principal objective in this article is to test for periodic structure in electricity spot prices and load data in order to establish whether the nature of the underlying periodicity permits us to competently predict the spot price and load far into the future. (1) As such, we are particularly concerned with the stability or predictability of the periodic structure of price and load time series data.
Our approach differs from the conventional conception of periodicity in the time series and signal processing literature which utilizes a deterministic periodicity (sinusoid) possibly embedded in additive noise. If the noise process is a symmetric or uniform noise process, then the periodicity will have a constant waveform, (Li and Hinich 2002, 1, Hinich and Wild 2005, 1557-1558).
Our approach also differs from the conventional approaches that have been used to model time series with changing periodic structure that can be classified as models with either 'seasonal' unit roots or 'season-dependent' parameters--see (Li and Hinich 2002, 1-2) for an overview of these two different approaches. (2)
In undertaking this, we employ a univariate approach, although, from both an economic and forecasting perspective, there is likely to be interest in broader questions concerning possible relationships between the spot price of electricity and electricity load and other covariates such as prices of primary commodities like coal and natural gas, which enter as input costs in electricity generation, economic activity and weather patterns. We see such wider investigations as complementary to the kind of statistical analysis undertaken here.
The paper is organized as follows. In Section 2 we discuss the Australian National Electricity Market (NEM). In Sections 3 and 4 we briefly outline the RMP model proposed by Hinich (2000) and Hinich and Wild (2001, 2005). In Section 5 we briefly discuss the data used and highlight some transformations that were made to the spot price electricity data in order to implement the RMP test. In section 6, we test for randomly modulated periodicity in half hourly electricity prices and MWh load demand. In the final section, we provide concluding comments.
2. AUSTRALIAN NATIONAL ELECTRICITY MARKET (NEM)
The electricity market as a whole encompasses both supply and demand side interactions. The Australian market for electricity is structured as a gross pool arrangement. This market structure is ideal for electricity because of its peculiar properties. First, electricity cannot be stored and supply must balance demand instantaneously through time. Second, because one unit of electricity is indistinguishable from all other units, it is not possible to determine from which generator the unit of electricity was produced (NEMMCO 2005, 4).
The electricity industry involves generation, transmission, distribution and retail sale activities. More than 90% of Australia's electricity production is generated from burning coal, gas and oil. In 2003, statistics relating to generation by fuel type indicated that approximately 58.5% of generation occurred by burning black coal, 25.9% by brown coal, 7.7% by natural gas, 7.6% by Hydro and 0.3% by oil products (NEMMCO 2005, 4).
The NEM commenced operation as a de-regulated wholesale market in New South Wales, Victoria, Queensland, the Australian Capital Territory (ACT) and South Australia in December 1998. In 2005, Tasmania joined as a sixth region. Operations are essentially based on six interconnected regions that broadly follow state boundaries. The market is extensive in scope with trade in electricity accounting for around $7 billion in 2003, meeting the demand of around 8 million consumers (NEMMCO 2005, 4).
The National Electricity Market Management Company Limited (NEMMCO) was established in 1996 to administer and manage the NEM. NEMMCO is a company under the Corporations law and operates on a break-even basis by recovering the costs of operating the NEM as well as it own operational costs by levying fees against market participants (NEMMCO 2005, 5). More generally, the structure of ownership of NEM infrastructure assets is complicated, with assets being owned and operated by both state governments (i.e. public ownership) and by private businesses (i.e. private ownership). In 2003, public (government) ownership encompassed around 64% of generation assets, 57% of transmission assets, 50% of distribution assets and 55% of retail assets (NEMMCO 2005, 5).
In Australia, the wholesale spot market for electricity is a key component of the NEM. The spot market can be viewed as being derived from a continuous auction market in which asks and bids are entered by generators and users of electricity to generate five minute (market clearing) dispatch prices that are broadcast to market participants in real time. Towards the end of the five-minute interval, market clearing is achieved through an economic dispatch algorithm that selects the cheapest available resource from the offers submitted by market participants to meet incremental changes in demand experienced by the real power system. The official trade (spot) prices and positions are determined by taking half hour averages of the five-minute dispatch prices and loads. The half hourly averaged prices are those received by generators and paid by purchasers of electricity (Outhred 2000, 3-4, NEMMCO 2005, 6-7).
Currently, NEM rules set a maximum spot price of $10,000 per megawatt hour. This is the maximum price that generators can bid into the market. This maximum price is also called the Value of Lost Load (VOLL) and is automatically activated whenever NEMMCO pursues load shedding in order to ensure that supply and demand balance and that the quality of supply meets pre-determined security and reliability standards (NEMMCO 2005, 6, 9).
There is also provision for price capping behavior associated with a Cumulative Price Threshold (CPT) that serves to cap potential financial risk in the NEM during periods of high sustained spot prices. This mechanism is triggered if the cumulative price in a single region over the preceding 336 trading intervals in a rolling seven-day period reaches some pre-specified threshold level. If this occurs, the maximum spot price is reduced from VOLL to an...
|
