Temperature Dynamics, Volatility, and the UK Demand for Natural Gas
By
Alec John Michael Horton
2010
A Dissertation presented in part consideration for the degree of MA Finance and Investments.
Abstract
The purpose of this paper is to consider how changes in temperature affect the volatility of the financial markets and overall demand for natural gas as released by the National Grid. Parts 1 to 3 of this paper provide an overview of the gas markets and the literature. Parts 4 and 5 provides the reader with an in depth analysis into temperature, demand and the financial markets. Part 4 finds a strong relationship between temperature and the demand for natural gas and clear evidence of seasonality in natural gas demand. Part 5 focuses on volatility and the financial markets. There is significant evidence that market volatility is greater during the winter months in comparison to the summer months. The market is also found to be largely inefficient, which is confirmed when testing the Efficient Market Hypothesis in part 5.1.
Acknowledgements
I would first like to thank my supervisor Dr Monica Giulietti who throughout the preparation of my dissertation has kept in constant communication and provided me with sound advice. This dissertation has been completed for Advisory Group AG, I would like to thank all the people at Advisory Group who helped me obtain my data and have remained in constant communication throughout this process, and I would also like to thank Advisory Group for several very enjoyable trips to Zurich throughout the writing of my thesis.
List of Figures and Tables......
List of Terms
1 Introduction
1.1 Theoretical view of the Natural Gas Markets
1.12 The Spot-Futures Parity and Efficient Market Hypothesis
1.2 What is Natural Gas?
1.3 Background on the UK Natural Gas Market
1.31 The UK Import Market
1.4 An overview of UK Weather
1.41 Climate Change
1.5 Why are the energy markets so volatile?
2 Literature Review
3 General Data Summary
4 The Demand for Natural Gas and Temperature
4.1 Data and Methodology
4.2 Model Specification
4.3 Estimation Results
4.4 National Grid Demand and Seasonality
5 The Financial Markets and Temperature
5.1 Testing the EMH
5.2 Which Market is more responsive to changes in temperature?
5.3 Data and Methodology
5.3 ARMA - Model Specification and Results
5.4 GARCH - Model Specification and Results
5.5 T-GARCH - Model Specification and Results
5.6 Seasonal Analysis
5.61 Monthly GARCH Analysis
5.62 Spot Price Volatility and Seasonality
6 Conclusions
7 Bibliography
7.1 References
7.2 Appendices
Figure 1 – Worldwide marketed Natural gas Energy Consumption (quadrillion Btu), IEO 2010.
Figure 2 – National Demand Index vs. London Daily Low Temperature
Figure 3 – Futures Price vs. Birmingham Daily Low, December 2009 – January 2010.
Figure 4 – National Grid Estimated Demand (Mcm),
Figure 5 – Price Behaviour and the Efficient Market Hypothesis, Brealey et al (2008)
Figure 6 – Hurricane’s Katrina and Rita, and their path toward the US mainland, August-September 2005. Wells, 2006
Figure 7 – Daily Natural Gas Production from the Gulf of Mexico following landfalls of Hurricanes Katrina and Rita. Wells, 2006
Figure 8 – UK Local Distribution Zones, National Grid 2009.
Figure 9 – ACF and PACF of Spot Price Returns.
Table 1 – Summary of Demand, Spot and Futures data.
Table 2 – Augmented Dickey Fuller Test for Unit Roots
Table 3 – Regression Coefficients and N-W S.E.
Table 4 – National Grid Demand and Seasonality Estimated Coefficients.
Table 5 – Evidence of Seasonal Demand for Natural Gas.
Table 6 – Spot and Futures Summary Statistics.
Table 7 – ADF, Jan 2006 – May 2010.
Table 8 – Portmanteau test for White Noise, Jan 2006 – May 2010.
Table 9 – Akaine’s Information Criterion – AR Component.
Table 10 - Akaine’s Information Criterion – MA Component.
Table 11 – Akaine’s Information Criterion - ARMA (p,q)
Table 12 – Test for ARCH effects
Table 13 – AIC for GARCH (p,q) models
Table 14 – Spot Returns, note that (α1 + β1=0.983<1).
Table 15 - Spot Returns, T-GARCH, note that (α1 - γ +β1=0.9502<1).
Table 16 – Monthly Dummy Variable GARCH (1,1) Analysis.
Table 17 – Spot Volatility vs. WSD and Seasons.
Table 18– Gas Sales and numbers of customers at regional and local authority level, 2007
Table 19 - Location of measurement stations and variable names, Bloomberg (2010).
Table 20 - List of Weather variables and definitions.
Table 21– London Summary Statistics.
Table 22– Birmingham Summary Statistics.
Table 23 –Glasgow Summary Statistics.
Table 24– Sunderland Summary Statistics.
Table 25- Manchester Summary Statistics.
Table 26– Nottingham Summary Statistics.
Table 27– Cardiff Summary Statistics.
Table 28– Southend Summary Statistics.
Table 29– Brighton Summary Statistics.
Table 30- Bristol Summary Statistics
Table 31 – Southampton Summary Statistics.
Table 32– Carlisle Summary Statistics.
Table 33- Heating Degree Days and Location correlations.
Table 34 – Test results for Heteroscedasticity for initial CLRM.
Table 35 – Durbin Watson tests of Serial Correlation.
List of Terms
95% Conf. / 95 percent confidence interval / GARCH (p,q) / Generalised Autoregressive Conditional HeteroscedasticityAIC / Aikaike Information Criterion / GWh / Gigawatt Hour
ADF / Augmented Dickey Fuller Test / HDD / Heating Degree Days
ACF / Autocorrelation Function / ICE / Intercontinental Exchange
AR (p) / Autoregressive / IEO / International Energy Outlook
ARCH (q) / Autoregressive Conditional Heteroscedasticity / LDZ / Local Distribution Zones
ARIMA (p,q) / Autoregressive Integrated Moving Average / MMcm / Million Cubic Meters
AUT / Autumn / MDV / Monthly Dummy Variable
BBL / Balgzand-Bacton-Line / MA (q) / Moving Average
Bcm / Billion Cubic Meters / NBP / National Balancing Point
LM / Breusch Godfrey Test / NTS / National Transmission System
BTU / British Thermal Units / N-W / Newey West
CLRM / Classical Linear Regression Model / OTC / Over-the-Counter
CLRM / Classical Linear Regression Model / PACF / Partial Autocorrelation Function
CHV / Composite Heating Degree Day Variable / GBP / Pound Sterling
CHV / Composite Weather Variable / R2 / R-Squared, measure of model fit
CWV / Composite Weather Variable / NBP97 / Short Term Flat NBP Trading Terms and Conditions 1997
CDD / Cooling Degree Days / SPR / Spot Prices (EMH)
UGASDEMD / Daily Actual National Grid Demand for Natural Gas / SPRN / Spring
MEAN / Daily Average Temperature / S.E. / Standard Error
NBPG1MON / Daily Front Month NBP Closing Prices (Futures) / SUM / Summer
HIGH / Daily High Temperature / UK / The United Kingdom
MIN / Daily Low Temperature / Thm / Therm
GA1NB / Daily Prompt NG Closing Prices (Spot) / T-GARCH (p,q) / Threshold Generalised Autoregressive Conditional Heteroscedasticity
DTI / Department of Trade and Industry / DD / Total Degree Days, HDD + CDD
DW / Durbin Watson Test / WSJ / Wall Street Journal
EMH / Efficient Market Hypothesis / WSD / Weather Surprise Dummy Variable
EIA / Energy Information Administration / WSV / Weather Surprise Variable
FPR / Futures Prices (EMH) / WINT / Winter
The first part of this paper will be devoted to an overview of the Natural Gas markets and key issues that will provide the foundation for analysis in future sections. Sections two and three will provide a summary of the literature and obtained data. Parts four and five will provide detailed analysis of both the level of demand and the financial markets for Natural gas, with a focus on the temperature and volatility. Within section six are conclusions to the analysis in sections four and five.
1 Introduction
1.1 Theoretical view of the Natural Gas Markets
When analysing the relationship between temperature and the price of commodities past literature has generally preferred the use of either Spot or Futures prices. This part of the paper will briefly discuss the Spot-Futures parity condition and its implications for analysis in further parts. Fama and French (1987) find that good spot-price data is not available for most commodities and prefer the use of futures, they state that futures are regulated through organised regulated exchanges and thus can be assumed to be a true reflection of the market. Spot prices data is released by reporting agencies such as Bloomberg, Reuters and Platt’s and these prices often differ across reporting agencies. Mu (2004) verifies Fama and French’s observations and also prefers the use of futures.
1.12 The Spot-Futures Parity and Efficient Market Hypothesis
The theoretically correct relationship between the spot and futures price is known as the Spot-Futures Parity, if this relationship fails to hold, arbitrage opportunities arise. There are essentially two ways to acquire a commodity such as Natural gas. Market participant can purchase the physical commodity today and store it, or can choose to take a long position in futures, these two strategies must have the same market determined costs. Commodities are physical goods and thus have different properties to financial assets, for example, a Natural Gas processing plant is not purchasing a futures contract to speculate but to consume. In absence of storage costs, the forward price of a commodity, such as Natural Gas is given by Equation 1. Where F0 is the Forward Price, S0 the Spot Price, r the risk-free rate of return and T the time period. This equation must hold to prevent risk free arbitrage profits.
Equation 1
If a term ‘u’ is introduced, which represents the present value of all the known storage costs that will be incurred over the contract period, absorbing funds, it follows Equation 1 that
Equation 2
There are also advantages to owning a physical commodity, if a Natural Gas processing plant is long futures and there is some unforeseen shock, such as an extremely cold winter, this will cause an increase in demand for gas, but, the plant cannot convert the futures contracts into physical delivery before contract maturity. The advantage to the plant of holding the physical commodity is difficult to quantify, but Hull (2002) prefers the term ‘convenience yield’, denoted by y and shown in Equation 3.
Equation 3
The convenience yield essentially represents market expectations in regards to the future availability of the commodity. The greater the likelihood of shortages for example, the higher the convenience yield. The difference between the futures price and the spot price is called the ‘Basis’, overtime the basis will be volatile but eventually converge. If today’s Futures price is equal to the expected spot price at maturity then;
Equation 4
Over an extended time period, in rational markets, expectations about futures spot price will adjust upward as often as downward. Telser (1958) finds that futures prices display no trend as they approach maturity and accepts the hypothesis that the futures price equals the expected spot price, Gray (1961) verified Telser’s findings and Dusak (1973) also supports Equation 4.
The spot-futures parity condition is consistent with the Efficient Market Hypothesis (EMH) (Fama, 1970) which states that the financial markets are informationally efficient. In an efficient market, new information is reflected instantly in commodity prices, which implies that the futures price is the optimal forecast of the spot price. No other topic has produced as many articles as the EMH in the area of finance[1]. The spot futures parity condition is based on the assumption that market participants are able to trade in the spot and futures markets at the same time, i.e. traders can utilize any spot/futures price differentials. If the EMH holds then price patterns are random and no system based on past market behaviour can earn excess returns.
Walls (1995) founds that the spot price of natural gas was co integrated with the futures price, that each price conveys the same information about the present and expected underlying value, that is, the markets are efficient and Shawky (2002) found that many of the characteristics of the electricity market can be viewed to be broadly consistent with efficient markets.
However, Chang (1985) found that ‘large wheat speculators’ possessed some superior forecasting ability and provides statistical evidence that is inconsistent with the hypothesis that commodity futures prices are unbiased estimates of the corresponding future spot prices, Houthakker (1957) also finds evidence of definite forecasting skill. In terms of the Natural Gas markets, Herbert (1993) was first to look at markets for US Natural Gas futures and found inefficiency in the market. Chinn et al (2005) found that futures prices were unbiased predictors of future spot prices, with the exception those in the natural gas markets at the 3-month horizon, and Mazighi (2003) rejects the hypothesis of efficiency in the futures markets for natural gas and concludes that forward prices are far from being optimal predictors of spot prices.
1.2 What is Natural Gas?
Natural gas is a colourless, odourless and shapeless fossil fuel found underground that is generated through the slow decomposition of ancient organic matter. This gas is generally found trapped in pockets of porous rock which is supported by impermeable rock, although natural gas is also found within oil reservoirs (Associated Natural Gas) or coal deposits (Coal-Bed methane). Natural gas is extracted through the use of wells drilled into the porous rock and is largely composed of methane, all other by-products must be removed at a processing plant before being moved through pipelines to the end consumer.
Natural gas is highly combustible and emits a great deal of energy when burned. Once delivered to homes it is used for a range of purposes, although in the UK it is primarily used to power central heating systems, boilers and gas powered ovens and increasingly Natural gas is being used to generate electricity. Consumers require space conditioning to create a comfortable living and working environment, electricity drives devices such as fans, air conditioners, chillers, cooling towers and electric boilers (Gellings, 2009) and energy use in buildings accounts for 53 percent of total electricity use (Harvey, 2010). Figure 1 illustrates that since 1990 Natural gas consumption has increased from 75.4 quadrillion Btu in 1990 to an estimated 162.3 quadrillion Btu in 2035 (International Energy Outlook 2010) and Harvey (2010) states that there are enough Natural gas reserves to last for 80-217 years depending on supply and demand approximations.
Figure 1 – Worldwide marketed Natural gas Energy Consumption (quadrillion Btu), IEO 2010.
1.3 Background on the UK Natural Gas Market
In the early 1980’s the UK gas industry began to liberalise and restructure[2], which began with the privatisation of British Gas in 1986 and the ‘demerger’ of its activities in 1991. Prior to the liberalisation of Britain’s energy markets British Gas and 14 regional public electricity suppliers had a monopoly to supply gas and electricity to every domestic energy consumer. Today the market is very competitive and the demand for Natural gas in mainland UK is categorized between Local Distribution Zones (LDZs), of which there are thirteen in the National Transmission System (NTS). Ofgas was created to ensure a smooth transition from a vertically integrated state owned monopolistic market to a competitive market in which consumer interests were protected.
The UK is one of the ‘big six’ major European gas markets along with Germany, Italy, France, Netherlands and Spain. The UK market is the largest volume market and is completely liberalised, because of this the UK market is also the most active, competitive and volatile gas trading market in Europe. Approximately 40% of the UK’s primary energy comes from gas, and there are large summer/winter swings due to ‘central heating’ demand, this is shown in Figure 2, which clearly shows that when temperatures are at their lowest, the demand for natural gas is highest and vice versa.
Figure 2 – National Demand Index vs. London Daily Low Temperature
The majority of gas in Europe is priced on a long-term contract, and a per country basis, short-term fluctuations tend to be due to traders trading out daily imbalances, these gas prices are based on ‘market values’ and competition at the National Balancing Point (NBP). In the UK market, the majority of demand for natural gas is during the winter months, summers are usually the lowest demand periods. However, abnormally hot periods can cause an increase in demand, as consumers demand electricity to cool their homes, this is usually found to be the case in the US market. Recent Natural Gas market volatility and the changeover of the UK from a net exporter to a net importer mean that security of supply is also a top priority for the UK.
1.31 The UK Import Market
The UK has historically been a net exporter of Natural Gas, however recently is became a net importer. In 2008 natural gas production was 70Billion cubic meters (Bcm) and consumption was 96Bcm (CIA World Fact Book, 2010). The UK is home to the most developed and liquid hub in Europe, the NBP, which started trading in 1996. The NBP is a virtual trading hub which covers the whole British transmission grid, it is a notional point which does not have an identifiable physical location. It is the trading point of UK short-term natural gas and is key to the price that domestic consumers pay and unlike the conventional European trading hubs, trades made at the NBP are not required to be balanced, there is no fixed fee for being out of balance. The NBP can be seen as the UK equivalent of the Henry Hub in the US, it is the pricing and delivery point for Natural Gas futures in the UK that are traded on the Intercontinental Exchange (ICE).
The UK has a growing import capacity, the majority of gas enters the NBP system passing through the five beach terminals in the North Sea, but there are also direct pipelines to Europe. Langeled is a crucial 1200km pipeline that brings Natural Gas from Norway, and is able to supply about 26Bcm per year, the Balgzand-Bacton-Line (BBL) from the Netherlands is also able to supply about 16Bcm per year. Zeebrugge is a physical trading point located in Belgium, this hub is joined to the Bacton terminal of the NBP through an interconnector pipeline that started operations in 1998 and has recently been upgraded to be able to supply 17Bcm per year. These three pipelines are able to meet a large part of the UK’s current demand. The original intention of the Zeebrugge pipeline for example, was to export gas from the UK North sea to Europe, but the flow of the Interconnector is often reversed to import gas to the UK market during winter months. Holz et al (2008) find that this increased pipeline import volume will compensate for the decline in UK domestic production.