Managing total corporate electricity/energy market risks

Electricity Journal

Volumn 11, Issue 8, 1998, Pages 36-46

  Henney, Alex ^a   Keers, Greg ^a  

^a NONE

Author keywords

[No Author keywords available]

Indexed keywords

POWER MARKETS;

BANKING INDUSTRY; CORPORATES; MARKET RISKS; TOOL KITS; VALUE AT RISK;

VALUE ENGINEERING;

EID: 4344649319     PISSN: 10406190     EISSN: None     Source Type: Journal    
DOI: 10.1016/S1040-6190(98)00083-9     Document Type: Article

References (15)

1. This section draws on several works, including Lillian Chew, MANAGING DERIVATIVE RISKS: THE USE AND ABUSE OF LEVERAGE (John Wiley & Sons, 1996),
2. an overview of the commercial occurrence and management of risk; Philippe Jorion, VALUE AT RISK (Irwin, 1997),
3. a textbook on VAR; and VAR: UNDERSTANDING AND APPLYING VALUE AT RISK (KPMG Risk Publications, 1997), a series of papers on financial applications of VAR. Also valuable was the latest source document on RiskMetrics, which is referenced in note 2.
4. RISKMETRICS: TECHNICAL DOCUMENT (Fourth Edition, New York, December 17, 1996), available on J.P. Morgan's Web page at http://www.jpmorgan.com/ Risk Management/RiskMetrics.
5. The table, cited in Chew (op. cit.), shows the significant differences in the VARs calculated in two ways for the equity and foreign exchange derivatives portfolios of a bank. The Basel Committee for Banking Supervision asked 15 banks from major G-10 countries to produce VAR numbers for a sample portfolio of approximately 350 positions and to test four different variants of the portfolio: balanced and unbalanced; and each of those with and without option positions. The returned results were widely dispersed because each firm used different correlation assumptions, volatility data, and treatment of options.
6. Dragana Pilipovic, ENERGY RISKS (McGraw-Hill, 1998) covers the quantitative ground.
7. Note that two other different meanings of the word "trading" are: (1) physical sale of surplus or purchase of deficit power; and (2) hedging to offset the risks of a portfolio of generation assets and/or of retailing contracts. Also, selling retail contracts is sometimes called retail "trading."
8. The "greeks" are as follows: the "delta" measures the sensitivity of the change of value of an instrument (e.g., option) or portfolio to change in either the spot price or the forward price, as appropriate. The delta is, thus, the partial derivative of portfolio value with respect to the market price of the underlying product (e.g., the market price of an electricity futures contract). Delta is related to position in that the product of an option contract delta (e.g., 0.4) and the contract size (e.g., 10 megawatts [MW]) gives an approximately equivalent position in a swap contract (e.g., 4 MW), the "delta equivalent" position. the "vega" risk measures the sensitivity of a change in value of an instrument or portfolio to a change in volatilities and is, thus, the partial derivative of portfolio value with respect to volatility. the "theta" measures the change in value of a portfolio with respect to time, as options tend to decay in value the closer they get to expiration. the "rho" measures the change in value of a portfolio with respect to a change in the discount rate. the "gamma" is the derivative of delta and is, thus, a "second-order" analytic. Portfolios that have non-zero gamma (e.g., options, caps, floors, and electricity retail and generation) exhibit non-linear behavior that causes inaccuracies in analytical or parametric VAR methods.
9. Taken to an extreme, electricity forward prices and their respective volatilities can be analyzed down to an hourly or half-hourly resolution. But collecting, processing, and forecasting market information to any useful degree of accuracy at this detailed resolution is impractical beyond an horizon of a few days. It is, therefore, common practice to group hours into time slots, which prima facie results in a degree of approximation when using such forward prices and volatilites for VAR calculations. In fact, the accuracy and precision of VAR calculations is generally improved by working with time slots because the reduced amount of data allows for higher-quality analysis through more accurate modeling of price uncertainties and distributions. The optimal time slots for risk analysis depends on regional market characteristics and the makeup of the particular portfolio being analyzed. In general, grouping hours that have similar price levels is largely consistent with accurate profit forecasting and risk quantification (e.g., weekday peak, weekday offpeak, and weekend).
10. Normally irrespective of the degree that output is financially contracted, but possibly affected by physical contractual commitments or market share motivations overriding those for operating profits.
11. Because of the dynamic nature, this modeling has to be done on spot prices and cannot be applied to forward prices, which have different stochastic behavior. For example, forward prices a year hence will not be as volatile as spot prices and will not spike up to extreme levels.
12. The way of eliminating electricity price and volume risk is to act as a marketing front for a generator who assumes those risks.
13. According to Jorion (op. cit.), a small part of Baring's portfolio losses were caused by trader Nick Leeson selling 35,000 puts and calls in equal volumes with similar strike prices. A delta-normal VAR calculation of this portfolio would have reported zero risk (the puts have negative deltas and the calls positive deltas), but a full-valuation simulation approach showed the real VAR to be $6.3 million.
14. Note that trading standard wholesale contracts to match the latest forecast position does not result in a "closed" position, since considerable risks remain due to volume uncertainties.
15. In no other commodity market are the reasons for an integrated analysis so compelling. For example, oil companies do not need to undertake market-to-market and VAR analyses that include petroleum stored at highway retail outlets, and a gold mining company is unlikely to include planned mine output in its market-to-market and VAR analyses. It is the lack of the buffering or insulating effects of storage that underpins the need for this approach with power portfolios.