0903 Financial Method


Energy prices, risk management and investment:
Interaction between policy-making and market behaviour


23rd March 2009

Department for Energy & Climate Change (DECC), 1 Victoria Street, London, SW1H 0ET

This short seminar is a follow-up to a UKERC residential workshop, held 9-10 July 2008 in Oxford|, to identify ways in which financial methods and modelling can tackle key challenges faced by the electric power industry. The policy-related findings of this workshop will be presented and discussed.

Draft Programme

DECC Presentation by Afzal Siddiqui

Abstracts of papers submitted for UKERC workshop:


"Policymaking Benefits and Limitations from Using Financial Methods and Modelling in Electricity Markets",

July 2008, Oxford.


Session I:  Empirical Analysis of Energy Derivatives

Volatility Transmission and Volatility Impulse Response Functions in European Electricity Forward Markets


The Relationship Between Day-Ahead and Forward Electricity Prices: Evidence from the UK  

Volatility Modelling with Value-at-Risk Applications for United Kingdom Natural Gas Futures Prices

Session II:  Investment Risk

Investment in Electricity generation: Why Policy Needs to Look Beyond Cost Modelling  

Private Sector Response to Carbon Price Uncertainty .

Efficient Investment Portfolios for the Swiss Electricity Supply Sector

Session III:  Real Options

Gas-Fired Power Plants:  Investment Timing, Operating Flexibility and CO2 Capture .

How to Proceed with the Thorium Nuclear Technology:  a Real Options Analysis 5


Session IV:  Carbon Capture and Sequestration

Using Financial Methods to Value Operating Flexibility: Opportunities for Improving Investment Decisions in Fossil-Fired Plants in the UK?

Issuing Tradable Capture Options with the Virtue of Financing Capture Ready and Assisting Policy Making in Carbon Capture and Storage .

Session V:  Demand Variability and Hedging

Optimal VaR Constrained Hedging of Fixed Price Load-Following Obligations in Competitive Electricity Markets

The Interplay Between Risk Aversion and Price Beliefs: a Key to Understanding Power Demand Elasticity  

Risk Management in Electricity Markets:  Hedging and Market Incompleteness 8

Session VI:  Financial Transmission Rights

Impact of Transmission Congestion on the Forward Risk Premium in Electricity Markets

Efficiency of Financial Transmission Rights Markets in Centrally Coordinated Periodic Auctions


Session 1 Empirical Analysis of Energy Derivatives

Volatility Transmission and Volatility Impulse Response Functions in European Electricity Forward Markets

Yannick Le Pen (Universit é de Nantes) and Benoît S é vi (Universit é d'Angers )


Our paper concerns the relevancy of volatility transmission models for risk management purpose. We consider the British, the Dutch and the German markets during the period March 2001 – June 2005. Our aim is first to detect volatility transmission between these three financial markets using standard multivariate GARCH analysis. If detected, the result is exploited by the mean of VIRF (Volatility Impulse Response Analysis) to measure the impact of a shock in one market on the volatility of another market. In words, the issue may be as follows: What is the dependence between European electricity markets in terms of volatility if measured using volatility proxies? When applied to our daily data (for base-load and peak-load), we observe that a shock has a high positive impact only if its size is large compared to the current level of volatility. The shocks' impacts are usually not persistent, which may be an indication of market efficiency. We then turn to the estimation of the density of our VIRF, in order to assess the robustness of the analysis. These results have interesting implications for market participants whose risk management policy is based on option prices which themselves depend on the volatility level.


The Relationship Between Day-Ahead and Forward Electricity Prices: Evidence from the UK

Fernando Oliveira and Dimitra Prompona (University of Warwick)


We analyze the empirical relationship between spot and future electricity prices, examining price returns and the realized forward premiums. We explore the issue of risk hedging in the British electricity market, comparing OLS, GARCH and cointegration as tools to estimate the hedge ratio. We show that OLS and co-integration give very similar results, whilst Garch results are inconsistent with the findings from OLS and cointegration. Furthermore, we analyze the dynamic properties and long-term relationship between spot and future electricity prices. We show that: a) day-ahead and forward prices are cointegrated; b) the market recovers fast from price shocks; c) disequilibria in the long-term relationship are corrected through movements in the day-ahead prices; d) Vector Error Correction model provides reliable long-term forecasts in an out-of-sample test.


Volatility Modelling with Value-at-Risk Applications for United Kingdom Natural Gas Futures Prices

Merrill Heddy (Baruch College), Joseph Onochie (Baruch College), Sjur Westgaard (Trondheim Business School), and Robert Yaffee (New York University)


Deregulation of many energy markets around the world has created the need for proper identification , definition, measurement, and  management of risk in these markets. In 2007, over 45% of electricity production in the United Kingdom employs natural gas as an input which suggests that volatility in the natural gas market in the U.K. would also reflect volatility in the electricity market. Sophisticated measures of risk have been developed for analysis of financial markets, but their application in energy markets is still being explored. Value-at-Risk (VaR) has recently been introduced as a technique of risk measurement and hedging in many areas of financial markets and measures the expected maximum loss over a specified horizon with a reported level of confidence. One major issue with the use of VaR has been the proper specification of the volatility. We therefore estimate and compare the volatility of UK natural gas returns using a variety of GARCH models with normal, t, and skewed t error distributions.  Another point of controversy is the assessment of accuracy of the VaR measure. We evaluate the accuracy of in-sample VaR estimation with simulation and the accuracy of out-of-sample VaR forecasts with measures of GARCH forecast evaluation. As an alternative coherent measure of risk, we also look at expected  shortfall to give us a better idea of the loss sustained from the  whole distribution.


Session II:  Investment Risk

Investment in Electricity generation: Why Policy Needs to Look Beyond Cost Modelling

Rob Gross, Will Blyth, and Phil Heptonstall (Imperial College)


Policy goals frequently depend upon investment in particular technologies, or categories of technology. Policy goals such as security of supply, or reducing CO2 emissions might favour nuclear power, coal with CO2 capture or renewable energy. Policy decisions are often informed by estimates of cost per unit of output (e.g. £/MWh) also known as levelised costs. Costs are often used to provide a 'ballpark' guide to the levels of support needed (if any) to encourage uptake of different technologies.  They can also help to indicate the cost of meeting public policy objectives, and whether there is a rationale for intervention (for example based on net welfare gains). Yet, because investment is undertaken by private companies, not governments, policy must be designed with the investment risks, not just technology costs, in mind.


Investment is driven by expected returns, in the light of a range of risks related to both costs and revenues. Revenue risks are not captured in estimates of cost or cost related risks.  An important category of revenue risks are associated with electricity price fluctuations. Exposure to price risks differs by technology. Low electricity prices represent a revenue risk to technologies that cannot affect electricity prices (nuclear, renewable and hydro plants). By contrast, 'price makers' that set marginal prices are, to an extent, able to pass fuel price increases through to consumers. They have an inherent 'hedge' against fuel and electricity price fluctuations. This paper explains why this is the case. Using assumptions from the DTI Energy Review (2006), the authors contrast the range of levelised costs estimated for different generating options with the spread of returns each is exposed to when electricity price fluctuations are factored in. 


Drawing on recent policy experiences in the renewable energy arena in the UK and elsewhere the authors provide an assessment of investment risk in policy effectiveness and consider how policy design can increase or ameliorate price risk. They discuss the circumstances under which policy goals might be best served by 'socialising' price risk, through fixed price policies.  The importance of increased and explicit attention to revenue risk in policymaking is discussed, including the value of factoring such risks into levelised costs through adjustments to discount rates.


Private Sector Response to Carbon Price Uncertainty

Karsten Neuhoff (University of Cambridge)


CO2 policies are increasingly, and with growing stringency, implemented in various countries. The private sector response to these policies will determine future emissions reductions. A robust evidence base and framework to understand this response is still lacking, but should be the basis for assessment of low carbon policy instruments and private sector investment options.


We analyse and characterise how the investment landscape was affected and is evolving under CO2 policy. In particular we focus on the following four research questions:


How do private sector investors represent CO2 policies in their investment choices?

How do private sector investors assess, quantify and represent uncertainties about CO2 prices and evolving carbon policies?

How do private sector investors hedge risks associated with uncertainties about CO2 prices and evolving carbon policies?

Do early moving private sector participants anticipate future market opportunities in the new economic environment created by carbon policies?


Previous work has pointed out that different sectors have fundamental different risk perceptions, risk assessments and means of risk management. Therefore any meaningful analysis will have to be sector specific. The research builds on a set of interviews first conducted in spring 2007. The insights provided by this work are the basis for further interviews to be conducted by March to allow for quantitative comparisons across sectors and a model representation.


The above paper provided basis for a joint paper between Karsten Neuhoff and Thomas Weber:


Wee examines the effects of firm-level innovation in carbon-abatement technologies on optimal cap-and-trade schemes with and without price controls. We characterize optimal cap-and-trade regulation with price cap and price floor, and compare it to the special cases of pure taxation and simple emissions cap. Innovation shifts the trade-off between price- vs. quantity-based instruments towards quantity-based emissions trading schemes. More specifically, a higher innovative effectiveness lowers the optimal emissions cap, and leads to looser price controls unless the marginal environmental damage cost is small. Because of the decrease in the emissions cap, innovation in abatement technologies can lead to a higher expected carbon price, so as to provide sufficient incentives for private R&D investments. The expected carbon price decreases, once innovative technologies are widely used.


Efficient Investment Portfolios for the Swiss Electricity Supply Sector

Reinhard Madlener ( Institute for Future Energy Consumer Needs and Behavior /E.ON Energy Research Center) and Christoph Wenk (University of Zürich)


In this paper we investigate existing and possible future power generation capacities in Switzerland from a risk-return perspective, using the Mean-Variance Portfolio Theory of Markowitz (1952). We study the various technologies in a rich and flexible framework, putting the main focus on technology-specific risks on the supply side, and the price level risk, which is largely influenced by the demand side. Further, we allow the risks tackled to have varying third and fourth moments, using econometric tools to find the most realistic distribution of the return. Portfolio optimization is based on the lifetime return on investment, obtained through a comprehensively defined Monte Carlo simulation that contains the predefined risks and is conducted both for a base-load and a peak-load portfolio. Our analysis starts from the current mix of power producing technologies and offers insights on how production should be shifted either to increase the expected portfolio return or, alternatively, to reduce its risk. Furthermore, our results suggest that when considering electricity production, the focus should not rest on a single technology or a single plant, but rather on the question how current capacity can be altered in order to increase total efficiency.


Session III:  Real Options

Gas-Fired Power Plants:  Investment Timing, Operating Flexibility and CO2 Capture

Stein-Erik Fleten (Norwegian University of Science and Technology) and Erkka Näsäkkälä (Helsinki University of Technology)


We analyze investments in gas-fired power plants based on stochastic electricity and natural gas prices. A simple but realistic two-factor model (Schwartz and Smith 2000) is used for price processes, enabling analysis of the value of operating flexibility, the opportunity to abandon the capital equipment, as well as finding thresholds for energy prices for which it is optimal to enter into the investment. We use the real options approach as outlined by Dixit and Pindyck (1994). Our case study uses representative power plant investment and operations data, and historical financial derivative prices from well-functioning energy markets. We assume that financial energy prices are efficient, in particular that they reflect the possibility of large power plants coming online. This means that we can model input and output prices as exogenous even though a large capacity increase may decrease electricity prices. We find that when the decision to build is considered, the abandonment option does not have significant value, whereas the operating flexibility and time-to-build option have significant effect on the building threshold. Further­more, the joint value of the operating flexibility and the abandonment option is much smaller than the sum of their separate values, because both are options to shut down. The effects of emission costs on the value of installing CO2 capture technology are also analyzed. Building a CO2 capture plant and piping CO2 off to permanent storage or in oil fields for increased recovery is a relatively expensive way of reducing greenhouse gas emissions.


How to Proceed with the Thorium Nuclear Technology:  a Real Options Analysis

Afzal Siddiqui (University College London) and Stein-Erik Fleten (Norwegian University of Science and Technology)


The advantage of thorium-fuelled nuclear power is that it limits the potential for spreading weapons-grade material and produces less long-lived nuclear waste than existing uranium-fuelled plants. However, there are a number of technical challenges that need to be overcome, and the current costs of initiating a thorium fuel cycle would be very high. We analyse how a government may proceed with a staged development of meeting electricity demand as fossil fuel sources are being phased out. The thorium technology is one possibility, where one would start a major research and development program as an intermediate step. Alternatively, the government could choose to deploy an existing renewable energy technology, and using the real options framework, we compare the two projects to provide policy implications on how one might proceed.



Session IV:  Carbon Capture and Sequestration

Using Financial Methods to Value Operating Flexibility: Opportunities for Improving Investment Decisions in Fossil-Fired Plants in the UK ?

Hannah Chalmers (University of Surrey), Matt Leach (University of Surrey), and Jon Gibbins (Imperial College)


In recent years, the value of flexible operations at coal-fired power plant operators in the UK has been significant as the previously unexpected 'dash-to-gas' combined with cheap gas prices meant that plants that had been designed as baseload generators were required to operate as mid-merit plants.  Volatility in fuel prices, uncertain future climate policy and ongoing changes in the mix of available electricity generators are expected to continue to play a significant role in determining how fossil-fired power plants are designed and used.  Yet, although there is considerable uncertainty over many key elements determining the operating patterns and hence profitability of particular power plants, it is expected that investors will have to make decisions about the next wave of UK power generation before the end of this decade if secure electricity supply is to be maintained through to 2020. 


This paper will discuss the potential for financial methods to improve policy design and investment decisions by allowing decision-makers to make quantified estimates of changes in risk profile associated with operational flexibility of different types of power plant.  A case study considering potential investment in the current UK market will be presented to highlight issues and methods that could be relevant both in the UK and globally.  In particular, the potential for techniques such as real options and portfolio analysis to provide insights that cannot be obtained with approaches that are currently dominant in economic modelling to inform UK energy policy will be discussed.


Issuing Tradable Capture Options with the Virtue of Financing Capture Ready and Assisting Policy Making in Carbon Capture and Storage
Xi Liang (University of Cambridge), David Reiner (Unive sity of Cambridge), Jon Gibbins(Imperial College), and Jia Li (Imperial College) xl260@cam.ac.uk

A Capture Option is an option contract which stipulates that the option holder has the right (but not the obligation) to exercise the contract to retrofit an underlying fossil fuel plant to capture CO2 on or before a fixed date. In developing a liquid market of Capture Options, key stakeholders such as plant owners, policymakers, and bankers will be better able to assess  the value of Capture Options which will encourage new fossil-fired plants to be developed as 'Capture Ready'. As a result, the Capture Option concept can improve financing schemes for Capture Ready investments and assist policymakers in formulating the policies for Carbon Capture and Storage (CCS). In a detailed case study based on data for one of the leading Chinese electricity generating companies, we assessed the benefit of a Capture Option and Capture Ready investment for a 600 MW supercritical pulverized coal fired power plant using Real Options Analysis (ROA) of a cash flow model with Monte-Carlo simulations. At an 8% discount rate, th e gross value of Capture Ready varies from CNY3m to CNY633m and the fair price of the Capture Option is was CNY113m and 1255m for two of the four scenarios analyzed; furthermore, Capture Ready investment reduces the probability of early closure by 46% and 62% in the two scenarios and pushes forward the mean retrofitting year by 2 and 5 years respectively.


Key policy-related points:

One of the weaknesses in current economic (and techno-economic) literature on CCS is that it generally doesn't take account of different operating modes that could allow operators to have flexibility to react to various changes in the operating environment for power plants (both expected and unexpected and over short and long timescales).


These operating modes are also generally not included in analysis by policy-makers and this could lead to poor design of regulations and/or incentives for CCS that could be very detrimental to meeting the challenging (although still achievable) targets for technology development and roll-out that have been included in a range of energy systems studies recently (e.g. IEA Energy Technology Perspectives and World Energy Outlook).


Methods developed in the financial economics literature could provide some more robust insights into plant value under uncertainty (including operator and investor perception of risk) but there are also limits to what we can expect any particular modelling technique to be able to tell us.


Session V:  Demand Variability and Hedging

Optimal VaR Constrained Hedging of Fixed Price Load-Following Obligations in Competitive Electricity Markets

Yumi Oum (Pacific Gas and Electric) and Shmuel Oren (University of California, Berkeley)


Load serving entities providing electricity to regulated customers have an obligation to serve load that is subject to systematic and random fluctuations at fixed prices. In some jurisdictions like New Jersey, such obligations are auctioned off annually to third parties that commit to serve a fixed percentage of the fluctuating load at a fixed energy price.  In either case the entity holding the load following obligation is exposed to the load variation and to a volatile wholesale spot market price which is correlated with the load level. Such double exposure to price and volume results in a net revenue exposure that is quadratic in price and cannot be adequately hedged with simple forward contracts whose payoff is linear in price. A fixed quantity forward contract cover, is likely to be short when the spot price is high and long when the spot price is low.   In this paper we develop a self-financed optimal hedging portfolios consisting of a risk free bond, a forward contract and a spectrum of call and put options with different strike prices.  The portfolio is designed to meet a value at risk (VaR) constraint on the net hedged revenue of an entity holding a fixed price load following obligation.  For distributions where the VaR is a function of the mean and standard deviations, which is monotonically decreasing in standard deviation and increasing in the mean, it has been shown that an optimal VAR constrained portfolio is on the efficient frontier with respect to a mean-variance portfolio selection criterion.   We exploit this property to derive the VaR constrained  hedging portfolio as an equivalent optimal mean-variance portfolio under particular distributional assumptions.  We verify the results by proving that indeed the distribution of the resulting hedged total net revenue under the mean-variance criterion meets the needed property to support the equivalence to a VaR constrained optimal hedge.  The results are illustrated through a numerical example

The Interplay Between Risk Aversion and Price Beliefs: a Key to Understanding Power Demand Elasticity

Sandro Sapio (University of Naples Parthenope/Sant'Anna School of Advanced Studies)


Exposing power consumers to real-time fluctuations is often suggested as a key step towards achieving highly responsive demand in power exchanges. The effectiveness of such a policy can however be hampered by risk aversion. This paper is a theoretical attempt to understand the consequences of risk aversion for the responsiveness of demand to price fluctuations in power exchanges. In the paper, it is assumed that the electricity consumption choice by industrial companies depends on both the forecasted mean and variance of the power price, obtained using information from previous market sessions. Demand is supposed to increase if the power price goes down and, at the same time, the variance goes up. However, some of the main forecasting models used by academics and practitioners imply that mean and variance move together. For this reason, the power demand responsiveness to price could turn out to be small even if consumers are exposed to real-time fluctuations.


Risk Management in Electricity Markets:  Hedging and Market Incompleteness

Bert Willems ( Tilburg University/KU Leuven ) and Joris Morbee (KU Leuven)


This paper aims at a better understanding of the origin of risk in electricity markets and studies how electricity firms should manage their risk exposure. The paper develops an equilibrium model of the electricity market with risk averse firms, and compares the aggregate welfare under different assumptions with respect to market completeness, i.e. the number and type of derivatives which are traded in the market.  We show that aggregate welfare in the market increases with the number of derivatives offered in a particular market, but most of the benefits are achieved with a rather limited set of derivatives. Forward prices are biased estimates of future spot prices; however, this bias does not depend on the number of other derivatives, which are traded in the market. This bias is eliminated when speculators (traders) are present in the market.


Session VI:  Financial Transmission Rights

Impact of Transmission Congestion on the Forward Risk Premium in Electricity Markets
Shi-Jie Deng (Georgia Institute of Technology) and Haibin Sun (Constellation Energy Group) deng@gatech.edu  

This work has been partially supported by National Science Foundation grants

We examine the optimal hedging and equilibrium pricing of electricity with the presence of both forwards and spot markets subject to transmission system constraints. We assume that an independent system operator (ISO) organizes a wholesale electric power market where the participants consist of risk-averse generation companies (GenCos) and load serving entities (LSEs). GenCos and LSEs transact in the forwards market prior to the opening of the spot market. Electricity spot and forward prices are determined by a set of market clearing rules which result from the interactions between GenCos, LSEs and the ISO. The decision problems of each market participant in the forwards and spot markets are formulated. An LSE serves its loads with full service obligation in offering electricity to end consumers at contracted prices. To reduce risk exposure to the purchasing of electricity in the spot market, the LSE may procure the majority of its expected load obligations from forward markets and cover the load deviation in the spot market as the forward markets are typically much less volatile than the spot market. A GenCo makes its profit-maximizing generation schedules by taking into consideration the generation cost and sales revenue from both forward and spot markets. The system operator ensures the balancing of GenCos' supply and LSEs' demand with the least cost. Consequently, the electricity prices are formed at all locations in the bulk power system, reflecting the marginal cost of meeting incremental loads at their respective locations given transmission and other system constraints. Given the day-ahead and spot markets' clearing rules set by the system operator and the fact that LSEs' and GenCo's maximize their risk-adjusted profits, we derive the forward and spot market equilibrium prices in which the transmission congestion information is incorporated. This model is then used for analyzing risk premium observed in the electricity day-ahead forward prices and shedding light on how the forward premium is affected by the market participants' risk preference as well as the transmission network constraints. We conduct an empirical analysis using a historical data set which includes hourly electricity dayahead and real-time prices, loads, and transmission congestion information obtained form the New York electric power markets for the period of February 2005 to August 2006. The results of our preliminary analysis demonstrate that the relationship between day-ahead forward price premiums and the transmission network congestion measures is statistically significant.


Efficiency of Financial Transmission Rights Markets in Centrally Coordinated Periodic Auctions

Seabron Adamson (CRA International), Thomas Noe (University of Oxford), and Geoffrey Parker (Tulane University)


As Europe continues to develop various regional approaches towards implicit and explicit market coupling and congestion management, electricity market design in the United States is increasingly dominated by locational marginal pricing (LMP) of energy and transmission. LMP markets are coupled with periodic auctions of financial transmission rights (FTRs) to hedge transmission price risks. The New York Independent System Operator (NYISO) market was one of the first markets implemented to use LMP and centralized auctions for managing the trading of transmission rights. While LMP designs offer considerable advantages, forward price discovery in these markets requires participants to form efficient expectations on spot congestion price differences. In this paper, we examine trends in the efficiency of the NYISO centrally-coordinated transmission auctions using a panel data set of over 9000 contracts over a five year period, using robust econometric techniques. We find that the efficiency of the early centralized NYISO transmission auctions was relatively low and that the rate of improvement has been less than expected. We also find that the persistence of forwardspot pricing differences cannot be easily explained through risk aversion. Our results suggest that the ability of market participants to form efficient price expectations is constrained by the availability of detailed information about grid characteristics and operations. Rules about grid information release are therefore a critical issue for policymakers to consider as they seek to implement effective transmission auctions and electricity markets.