What Is It Worth? Application of Real Options Theory to the Valuation of Generation Assets

Electricity Journal

Volumn 14, Issue 8, 2001, Pages 40-51

  Frayer, Julia ^a   Uludere, Nazli Z ^b  

^a London Economics Inc   (United States)

^b NONE

Author keywords

[No Author keywords available]

Indexed keywords

COMMERCE;

COAL FIRED PLANTS; EFFICIENT PLANTS; GAS-FIRED; GENERATION ASSETS; MARKET PRICE; OPTIONS-BASED VALUATION; PLANT OPERATIONS; REAL OPTION THEORY; REAL OPTIONS; REGIONAL MARKETS;

COST BENEFIT ANALYSIS;

EID: 0346485713     PISSN: 10406190     EISSN: None     Source Type: Journal    
DOI: 10.1016/S1040-6190(01)00237-8     Document Type: Article

References (17)

1. All figures are in present value terms, discounted using the risk-free rate of 5 percent. The risk-free rate is necessary to maintain the risk-neutral valuation (RNV) assumption in our real options valuation framework.
2. In this simplified analysis, the modeling horizon was limited to 10 years (from 2001 to 2010) and cash flows were discounted without assigning any terminal value.
3. For the scope of this article, the useful lives of the assets were assumed to equal our modeling horizon of 10 years.
4. The spark-spread principle assumes full flexibility in operations and in fuel use; however, a peaking plant may be unable to capture this full flexibility due to the structure of its fuel contract. Such limiting factors can, however, be incorporated into the valuation.
5. F. Black, The Pricing of Commodity Contracts, J. FIN. ECON., Sept. 1976, at 167-79; and F. Black and M. Scholes, The Pricing of Options and Corporate Liabilities, J. POL. ECON., May-June 1973, at 637-59
6. F. Black, The Pricing of Commodity Contracts, J. FIN. ECON., Sept. 1976, at 167- 79; and F. Black and M. Scholes, The Pricing of Options and Corporate Liabilities, J. POL. ECON., May-June 1973, at 637-59
7. Estimating the price volatility of commodities, such as electricity and gas, by simply taking the sample standard deviation, would tend to overestimate volatility by disregarding mean reversion in such commodity prices.
8. Although we have utilized volatilities that are specific to peak and off-peak time periods, the model is capable of incorporating hour-specific volatility parameters.
9. NYMEX's electricity futures trade only 18 to 36 months in advance.
10. Though such models tend to underestimate the volatility in hourly electricity prices, they are good predictors of average trends. Thus, we do not use the production-cost-based models for determining the volatility parameters, but we do use them to construct the annual average price paths.
11. Such analysis can even be done for markets with no market data. This would require careful selection of a proxy region, one which would fundamentally exhibit a similar hourly price profile, due to similarity in system characteristics, shape of load profile, and supply/demand balance.
12. Gas prices are from Oilnergy; coal prices are from Natural Gas Week.
13. Gas prices are from Oilnergy; coal prices are from Natural Gas Week.
14. The intuition for the use of natural logarithm of prices is very simple: The options pricing models need to make assumptions regarding the underlying distribution of the price of the asset that is being valued. A common assumption is to assume that prices are normally distributed; however, since prices are never below zero, the normal distribution is not a correct characterization. The lognormal distribution, on the other hand, allows for only positive outcomes (in line with prices above $0) and is still applicable within the basic Black-Scholes framework (the natural log of prices is normally distributed).
15. The dummy variable regression essentially deseasonalizes the electric and fuel price series. First-order autoregression then corrects for mean reversion. This procedure is a generally accepted methodology for estimating volatility for energy commodities.
16. We discounted all values at a risk-free rate of 5 percent, which is consistent with the real options theoretical framework. Since liquid markets are often not available to hedge all risks, real options modeling methodology requires the stronger assumption of risk-neutral firms.
17. The flexible DCF value does take into account any costs of inflexibility, such as start and stop costs.