Risk management techniques and the associated language can be quite impenetrable for compliance managers in the commodity industry. This article serves as a starting point for those new to the topic, and demonstrates that energy risk management tools are essentially a relatively simple adaptation from the financial services industry. This basic tool has been the starting point for more advanced solutions, yet the fundamental ideas remain the same.
Energy companies have traditionally turned to the financial services industry in search of risk management tools and processes for their trading departments. A popular metric adopted from the financial services world has been value at risk. VaR is a probabilistic measure for the maximum estimated trading loss with a certain confidence (see chart one below). VaR is used extensively in the financial services world to estimate trading risk. There is, however, an ongoing debate about what constitutes the best risk management framework for oil and energy markets.
PA Consulting Group's experience from working with some of the world's largest energy firms indicates that risk management frameworks for the energy market should consist of an enhanced version of VaR, combined with a set of supporting tools. In broad terms, VaR is the most appropriate tool for measuring trading risk currently available. Furthermore, our experience shows that, although the basic definition of VaR has some limitations, these can be mitigated though some specific enhancements. Finally, a complete risk management framework requires some supplementary tools.
VaR is a good starting point
Market risk arises from the exposure of a position to market price movements. Determining the exposure of a position is a relatively simple task. (Structural positions such as long-term oil field production, option contracts and structures and diverse portfolios can make exposure calculations less straightforward.) Starting with nothing, the purchaser of 100 barrels of crude oil is 'long' 100 barrels. The point of a risk measure is to give managers an idea of the size of potential financial losses they may incur, based on the exposure of their positions and likely future market price movements.
A risk metric must be:
- A financial metric: the risk metric should be directly used to limit potential losses and manage financial performance.
- A probabilistic metric: by definition, when quantifying risk we are quantifying uncertainty. Mathematical probability theory is the most appropriate framework for this task.
- Capable of capturing portfolio effects. A good risk metric must capture the fact that multiple positions within a portfolio tend to have reinforcing or offsetting effects on risk.
VaR fulfils all of the above criteria, because:
- VaR reflects the change in value of the portfolio as a result of potential adverse movements in prices. It can therefore be used to limit potential losses and hence manage financial performance.
- VaR is a probabilistic metric. The VaR of a position is derived from the probability distribution of its future value, which in turn is derived from a probability distribution of future prices based on expected price volatilities. It is expressed as the size of loss that will not be exceeded at a certain confidence interval (normally 95 per cent or 99 per cent).
- VaR captures portfolio effects by incorporating the correlations between the movements of prices in different instruments and markets.
Despite these features, VaR measures have certain shortcomings, in particular in respect to energy markets.
Adopting an 'enhanced' version of VaR can mitigate these shortcomings.
VaR's weaknesses can be mitigated
VaR has certain limitations in the context of energy markets.
VaR assumes positions can be exited rapidly. This means that VaR can be poor at modelling long- and medium-term exposures, a particular problem in energy markets where physical positions can be very long-term and portfolios will be managed depending on price movement.
Energy supply contracts can run for years, if not decades, and physical positions are often slow or difficult to exit. For example, it takes a long time to renegotiate a long-term wholesale gas deal or to divest from a power plant. Similarly, many such contracts contain embedded optionality and so the net volumes will change as prices move. These two issues challenge the underlying assumption of VaR that market conditions and the shape of the portfolio will remain as currently observed during the time of the analysis.
To offset this, VaR models must be able to use different closeout assumptions in various markets, according to liquidity and the types of positions that the trading organisation holds. In addition, the scenario modelling undertaken must take into account how the volumes and exposures in the portfolio will react under changing prices as a result of optionality.
In energy markets where positions are held to delivery and supply or demand volume risk may result in imbalance payments, particular adaptations are required. Measures that are designed to cope with this are often termed profit at risk.
VaR is poor in modelling extreme price scenarios. Such scenarios tend to be more frequent in energy markets, which tend to be more sensitive to current events, than in financial markets. Natural disasters such as Hurricane Katrina in September 2005, terrorist attacks and international relations can result in energy price spikes and can have a significant short-term impact on volatility.
Such price spikes typically lie in the tails of the value probability distributions used in the calculation of VaR, hence VaR's accuracy at predicting such events will be poor. (See chart two below.)
This problem can be mitigated by the use of stress testing. Stress test models cover very unlikely but severe scenarios, such as large and counterparty defaults or extreme price volatility due to large-scale supply disruption. Stress tests therefore allow traders to explore what happens at market price extremes, and provide more information about the kind of risks that reside in the tails of the probability distributions.
By mitigating these two problems we produce a robust version VaR, which should then be supported by further controls.
Additional tools complete the risk management framework
In addition to the scenario-based stress tests described above, energy traders should supplement VaR with additional tools and controls such as:
- Marked-to-market performance. Daily marked-to-market reporting processes for the valuation of traders' books and calculation of open positions can help trading managers track performance, spot positions that are starting to make losses and get a better feel for the risk being taken in forward markets. This information should be combined with VaR to build a clearer picture of risk.
- Stop losses on strategies and portfolios. Stops place a limit on potential losses by forcing traders to exit positions that have lost a defined amount of value. Although traders will typically overshoot the stop loss limit due to market illiquidity under extreme conditions, such limits can be an effective way of curbing potential losses.
- Back testing of VaR models. By comparing historical mark to market to historical VaR data, trading organisations can determine and refine the accuracy of their VaR modelling assumptions. Such a process continuously improves the quality of VaR estimates.
VaR is therefore the obvious starting point for energy trading risk management. Although it has certain limitations, enhancing the sophistication and application of the VaR can mitigate those shortcomings. Combining this with specific additional controls and tools will produce an appropriate and comprehensive energy risk management framework.
Chart one: Probability distribution of the value of a position after three days
An example of VaR: Consider a NYMEX heating oil futures position, whose value has the probability distribution shown above. When we say that the position has a three-day, 95 per cent VaR of $5,000,000 as shown above, we mean that we are 95 per cent confident that the position's value will not decrease by more than $5,000,000 over the next three days. However, there is a five per cent chance that losses may exceed $5,000,000, and in extreme scenarios they can be significantly larger.
Chart two: Brent month ahead contracts - distribution of log daily price changes between May 16 2002 and September 23 2005
Brent crude oil month ahead forward contract price changes: historical price analysis indicates that 'tail' events are more frequent in energy markets than standard financial engineering price assumptions indicate.
This article serves as a starting point for those new to the topic, and demonstrates that energy risk management tools are essentially a relatively simple adaptation from the financial services industry. This basic tool has been the starting point for more advanced solutions, yet the fundamental ideas remain the same.
For further information, please go to:
http://www.paconsulting.com/industries/energy/risk_management/risk_reporting/
