Some in Congress would link customer choice with a portfolio standard. How would that play in a wholesale power market where gas turbines rule the roost?
By Michael C. Brower and Brian Parsons
WHAT KINDS OF POWER PLANTS WILL
get built in a deregulated electric industry? If recent history offers any guide, utilities and independent power companies will succumb to the traditional wisdom and invest in gas-fired combustion turbines and combined-cycle plants. Sound reasons may exist for doing so. The plants are less expensive than conventional steam plants. They put less capital at risk. Combined-cycle units have become very efficient, further reducing the cost. Moreover, natural gas prices are low and seem destined to stay that way for years.
But these attractive characteristics of gas technology should not blind power companies to the value of a diverse base of generating resources. For instance, what happens if gas prices rise unexpectedly? Also, the United States has been inching closer to making binding commitments to reduce greenhouse gas emissions, which could increase costs for owners of all fossil-fuel-burning plants. Lastly, electricity demand may become much more difficult to predict in a competitive market. That could make all large power plants (em even gas plants (em risky investments.
Diversifying the resource base by investing in renewable energy sources (em and in particular, wind plants (em offers an appealing strategy for managing such risks. Wind plants consume no fuel, can be built in relatively small increments with short construction lead time and generate no air pollution or greenhouse gases. These benefits, in fact, are among the main motivations for a provision of Rep. Schaefer's electricity restructuring bill (H.R. 655). The provision, the Renewable Portfolio Standard (RPS), requires the electric industry to meet rising targets for renewable energy use: 2 percent in 2000 rising to 4 percent in 2010.
We performed a study comparing the effects of investing in a gas-fired versus a wind power plant and learned that wind energy investments can cut risk in a variety of situations.
A Role for Renewables
The logical question to ask is how much risk-reduction value can wind energy provide. Will it be enough to tip the balance against less-expensive, gas-fired power plants? Or will it prove to be a minor factor?
We addressed these questions by comparing the effects of investing in one of two resource options, a 400-MW, gas-fired, combined-cycle plant and a 1600-MW wind power plant. (We assumed that the two plants had equal firm capacity.) The case study utility was Texas Utilities Electric, a large investor-owned company serving an area with abundant windy land. The uncertain inputs included fuel prices, environmental regulations (specifically, CO2 controls), wind plant output, conventional plant availability and load growth. We considered both the change in expected cash flow, and the value of changes in the risk to cash flow (measured by variance), resulting from these uncertainties.
Three different market scenarios were examined: 1) traditional regulation; 2) an unregulated spot market or power pool (modeled after the U.K. Pool); and 3) an unregulated market in which power is traded through fixed-price contracts of varying duration.
Our findings suggest that there exists a substantial risk-reduction benefit of wind energy under a range of conditions. With traditional regulation, the benefit for ratepayers is around $3.40 per megawatt-hour to $7.80/MWh. Company shareholders get no benefit because most risks under this system of regulation are passed on to the customers. In the unregulated scenarios, risks are divided differently between shareholders and consumers so the benefits of the wind investment are divided differently as well. In a power pool setting, we found that utility shareholders would receive the equivalent of an extra return on equity (ROE) of 1 to 1.5 percentage points, but consumers would end up worse off. Fixed-price contracts of a one- to five-year duration appear to share risks more evenly between customers and shareholders, giving both groups a modest incentive to choose wind.
Whether risk considerations tip the balance in favor of wind and other renewable resources must be considered case by case. Electricity from new gas plants is undoubtedly inexpensive, and substantial greenhouse gas restrictions or taxes may arise some
distance down the road. Moreover, other risk-management options, such as long-term fuel contracts, may be available. Nevertheless, our TU study showed that risk considerations could transform the wind plant from a clear loser into nearly a break-even proposition for that utility.
Effects on Risk
First, we review the base (fossil) and alternate (wind) plans under expected conditions, that is, allowing no deviations in fuel prices, load growth, environmental costs or plant availability (see Table 1). The cost streams are discounted at two different discount rates, the utility's weighted average cost of capital (WACC), 9.64 percent, and the presumed risk-free discount rate, 7.5 percent. In either case, the forced addition of the wind plant in 2003 increases revenues and net income and decreases costs. (Note that net income equals revenue minus cost.) The higher net income is necessary to compensate company shareholders for their larger investment in the wind plant, as is evident from the fact that the return on equity (ROE) in both cases is the same.
If risks and environmental externalities were ignored, the gas-fired, combined-cycle unit would be the preferred choice, since it is approximately $300 million less expensive for ratepayers. Taking risk factors into account can change this picture, however, depending on how the market allocates risk between customers and shareholders.
Traditional Regulation. In this scenario, electricity prices are not market-determined but set by the regulatory system to achieve a target rate of return on equity for TU Electric's stockholders. Changes in fuel prices and environmental costs are passed on to customers through a fuel-cost adjustment
to the base electricity rate.
Consequently, it can be expected that shareholders will have the least to gain from investing in wind as a risk-management strategy, whereas ratepayers will have the most to gain.
This result can be illustrated in a table that shows the expected present value and standard deviation of revenues, costs, net income and average return on equity for both the gas and wind cases and the differences between them. (See the first block of data in Table 2. High-risk environmental and fuel cost distributions are assumed.) The mean present value of revenues in the regulated scenario is $21 million greater with wind than without wind, suggesting that this case is still likely to be slightly more expensive for ratepayers, despite the possibility of CO2 regulation. However, the standard deviation in revenues is $464 million less, indicating that the wind investment is much less risky. By contrast, the mean return on equity is virtually the same in both cases.
Different views of the data provide additional insights into the effects of replacing the gas-fired plant with the wind plant. (Figure 1 shows a scatter plot of the difference in present-value revenues for the wind and gas cases in the regulated market scenario.) It is important to note that when the present value of revenue requirement in the gas case is high, the wind case tends to be less expensive than the gas case (points fall in the lower half the chart). In turn, when the present value of revenue requirement is low, the converse is true. (Note: This graphically illustrates the point that wind plants can act as an insurance policy or hedging strategy against fossil-fuel risks.)
Yet another view of the data (see Figure 2) shows the differences between the mean revenues and standard deviations of the gas and wind cases for each year of the study period. As expected, the wind case starts out more expensive than the gas case on average, but then becomes less expensive as fuel prices rise and the higher wind plant capital investment is paid off. In every year but the first, the standard deviation for the wind case is lower than that of the gas case by amounts ranging up to $150 million.
Poolco; Bilateral Contracts. The unregulated market is more complicated to model than the regulated market. The risks seen by the utility and its customers depend on many factors, such as the nature and degree of competition, corporate structures, the role of regulation, the design and functioning of the power pool, and the contractual relationships between the utility company and its
customers and fuel suppliers. We cannot incorporate all such factors into the model. Instead, we consider two scenarios that illustrate a plausible range of sensitivity to risk: a power pool (Poolco) scenario and a fixed-price contract (bilateral) scenario.
The critical difference between the two scenarios is that, in the power pool, TU Electric's plants compete against comparable fossil, nuclear, and renewable plants based on short-term variable
operating costs. Capacity payments are proportional to loss-of-load probability, as in the U.K. Pool. In the fixed-price contract scenario, the price of power is fixed for periods ranging from one to five years. In both cases, the capacity build decisions are assumed the same as in the regulated market scenario. The results can be summarized in a table. (See the bottom two blocks of data in Table 2.) It is important to note, first, that the power pool scenario poses greater risks for both customers and company shareholders than the fixed-price contract
scenario (i.e., the standard deviations are larger), whether wind is present or not. The reason is that the capacity payments in the power pool are highly volatile, as they depend on loss-of-load probability, which fluctuates greatly with peak load.
Moreover, the effect of substituting wind for gas varies strikingly between the two unregulated market scenarios. In the power pool scenario, the addition of wind appears to increase the standard deviation of revenues, but it decreases the standard deviation of the return on equity. The expected revenues, net income and return on equity are all somewhat lower with wind, to the benefit of electricity consumers but to the detriment of company shareholders. The main reason is that the wind plant slightly reduces the amount of high-cost fossil generation needed to supply loads at the margin and therefore reduces the variable portion of the electricity price. The results of the contract scenario, on the other hand, closely resemble those of the regulated market scenario. The main difference is the reduction in the standard deviation of return on equity resulting from the wind addition, which is accompanied by a slight increase in the mean ROE.
Valuing Risk Reduction
A critical issue in interpreting the results of a study like this one is estimating the value of changes in risk either for customers or utility company shareholders. There is, first, the possibility of a change in the expected, or mean, outcome, which occurs if the probability distributions of the input parameters are skewed in some fashion. In our study, the only such skewed distribution is that of environmental regulatory costs, which we believe are far more likely to increase than to decrease. The effect of this bias is easy to account for, and we already see it in the difference in mean revenues between the gas and wind cases in the regulated market scenario. (In Table 1, with no variations in the input parameters, the difference in mean revenues is $308 million, but in Table 2 it is $21 million. Thus, one can say that accounting for high
environmental regulatory risks reduces the mean revenues of the wind case compared with the gas case by $287 million.)
More challenging is the problem of assigning a value to changes in the variability of a cash flow. This is accomplished in decision analysis by calculating a risk premium, which is proportional to the
variance (or standard deviation squared) of some cash flow. This approach derives from expected utility theory. The certainty equivalent of the cash flow, which is the amount it is worth to a decision maker absent any risks, combines the mean with the risk premium in the equation:
where a is known as the risk aversion coefficient. In future cash flows, the certainty equivalent can be converted to a present value by discounting at a suitable risk-free discount rate.
The risk aversion coefficient can be measured directly by surveying the opinions or observing the investment behavior of the key decision makers or stakeholders. Absent such information, decision analysts generally assume it is approximately equal to the reciprocal of one to two times expected income. In this study, we assume, from the perspective of ratepayers, the risk aversion coefficient equals the reciprocal of 1.25 times revenues, whereas from the perspective of shareholders, it equals the reciprocal of 1.25 times expected return on equity. %n1%n
When the equation is applied to the annual means and standard deviations calculated by the model, and the wind and gas cases are compared, the result is an estimate of the total risk-reduction benefit of wind energy. (See Table 3. Note: The figures in this table have been divided by the amount of energy displaced in the substitution of the wind plant for the gas plant.) The total benefit is made up of two components, a change in mean revenues (due in our study entirely to environmental regulatory risks), and a change in the risk premium. Together, they show the consequences of allowing for risks in the comparison of the two resource options.
In the regulated market scenario, the benefit to the utility customer is equivalent in real levelized terms to $3.4/MWh to $7.8/MWh. The benefit is slightly smaller in the fixed-price contract scenario and actually negative in the power pool scenario. This means the customer is slightly worse off with wind in the power pool mix. The explanation for this last effect is unclear, but is likely connected to the way wind energy affects the dispatch of high-cost fossil-fuel plants operating at the margin.
The risk benefits from the shareholder's perspective are the
mirror image of those from the customer's perspective. In the power pool scenario, shareholders receive a major risk benefit from the wind plant that is equivalent to an extra return on equity of 1.2 to 1.5 percentage points. In the contract scenario, the benefit to shareholders appears smaller but still substantial (em 0.2 to 0.5 percentage points. As already noted, there is little or no risk benefit for shareholders in the regulated market scenario.
For the most part, accounting for risk points to benefits in adding wind energy to the fuel mix. The greater uncertainty in the annual average availability of wind plants compared with conventional plants does not offset the benefits of reduced exposure to fuel-price and environmental regulatory risks. This finding does not mean wind plants should always be selected, of course. In today's market, natural gas is so inexpensive that even considering risk factors it is still often the cheapest investment. However, risk could become a decisive factor when the cost differential between resources falls below $5/MWh.
Risks are distributed much differently in a regulated market than in an unregulated market, however, a fact that also affects the decision reached. In a regulated market, utility company shareholders see few of the risks of fossil fuels and so have little incentive to invest in risk-mitigation options such as wind power. This difference may help explain why many utilities have not eagerly embraced wind and other renewable resources in the past. An unregulated market may provide greater incentive for utility investment in wind energy based on risk considerations. Although this incentive is theoretically largest in a power pool, the extreme volatility of prices in such a market may serve to mask the incentive to a considerable degree. A market dominated by fixed-price contracts may be most favorable to wind, as the risk benefits will then be distributed more or less evenly between customers and utility company shareholders, giving both a modest incentive to go with wind. t
Michael C. Brower is president of Brower & Co., Andover. Mass. Brian Parsons is project manager for wind applications at the National Wind Technology Center, National Renewable Energy Laboratory, Golden, Colorado. The authors were assisted in this research by Kevin Bell, Convergence Research, Seattle; Peter Spinney, Charles River Associates, Boston; and Stephen Bernow and Max Duckworth, Tellus Institute, Boston. The authors gratefully acknowledge the support of the U.S. DOE for this research (NREL Subcontract No. TNA-5-15100-01).
1Support for these assumptions is provided in Jonathan M. Jacobs and Thomas E. Huntley, Pacific Gas and Electric Company, "Valuation of Fuel Diversity," Submitted for Hearings before the California Energy Commission (February/March 1992). Table 1. Gas vs. Wind
(Expected Risks (em Fuel Price, Load Growth, Availability, Environmental Compliance) %n2%n
Parameter Discounted at WACC %n3%n Discounted at Risk-Free Rate
Base Plan %n4%n 2 Alternate Plan %n5%n Change Base Plan Alternate Plan Change
PV of Revenues ($) 69,737 70,059 322 83,598 83,906 308
PV of Costs ($) 63,817 63,685 -132 76,674 76,469 -205
PV of Net Income ($) 5,920 6,374 454 6,924 7,437 513
Average ROE (%) 10.76 10.76 0.00 10.76 10.76 0.00
2Under expected conditions, i.e., no uncertainty in any variables.
3WACC - Weighted average cost of capital
4Base plan - 400-MW, gas-fired, combined-cycle plant
5Alternate plan - 1600-MW wind power plant
Table 2. Gas vs. Wind
High-Risk Assumptions (em by Regulatory Scenario)
Scenario Revenues ($) Costs ($) Net Income ($) ROE (%)
Mean Std. Dev. Mean Std. Dev. Mean Std. Rev. Mean Std. Dev.
Regulated Market Gas 91,159 17,503 83,629 16,337 7,530 1,843 10.70 1.14
Wind 91,180 17,039 83,154 15,846 8,026 1,857 10.71 1.11
Change 21 -464 -474 -492 496 14 0.00 -0.03
Unregulated Market Gas 100,270 23,705 88,111 17,973 12,159 7,379 21.80 11.32
Power Pool Wind 100,053 23,759 87,595 17,667 12,459 7,523 21.50 11.01
Change -216 55 -516 -306 300 144 -0.31 -0.31
Unregulated Market Gas 91,902 16,143 84,079 15,954 7,823 2,098 11.01 2.68
Fixed-Price Contracts Wind 91,984 15,755 83,609 15,474 8,375 2,085 11.14 2.47
Change 82 -388 -470 -480 552 -13 0.13 -0.21
*All figures are present values over 20 years (2003-2022) discounted at 7.5 percent, converted to 1996 dollars. Revenue and net income are in millions of dollars. Standard deviations reflect variations between iterations, not between years.
Technical Appendix Overview of the Model The model we used, Strategic Resources Planning (SRP) model, generates resource expansion plans and estimates capital and operating expenses for TU over a 20-year period. The resource plans and their estimated costs differ with each draw of uncertain inputs. For example, if gas prices increase sharply compared with coal prices in a particular draw, the model may select coal or wind instead of gas in its build decisions. The results of all the simulations are collected and presented as both an expected value and standard deviation of any indicator of interest (such as the present value of future revenue requirements or annual net income). Simulations are run until statistical errors fall to an acceptable level; this usually takes about 300 runs. The wind and gas scenarios are run simultaneously, using the same uncertain inputs, resulting in a very precise determination of the differences between them. It should be noted that the model does not consider the value of modular generating resources or option value (em the ability of a company to put off an investment until more information (on fuel prices, environmental regulations and other potential risks) is available. The capital costs and operating characteristics of new and existing fossil plants are based on TU Electric data. The wind capital cost is assumed to decrease from $908/kW in 1996 to $845/kW in 2003. Fuel Price Trends and Variations. Fossil fuel purchase is one of TU's largest expenses, constituting about $1.2 billion out of total operating revenues of $5.6 billion in 1994. It is also one of the most variable and unpredictable. TU's recent filings for the 1995 Integrated Resource Plan show an expected real growth rate in gas prices of 1.9 percent per year from 1994 to 2014, with a high rate of 2.8 percent and low rate of 0.5 percent. Little change is expected in coal or lignite. In the SRP model, it is assumed fuel price variations follow a random-walk process, with an adjustment for temporary price "shocks" from weather and supply and other factors. In a random walk, each annual price change establishes a new starting point to calculate the next year's price. Price shocks are assumed to disappear after one year. The random changes are drawn from a normal (bell-shaped) distribution. The mean of the distribution is TU's median price forecasts: 1.9 percent for gas and zero for coal. Two cases of standard deviation are considered. In the high-risk case, we assume future prices will reflect the volatility of the past 25 years. This implies, for gas, an overall standard deviation of 12 percent in the random walk and 10 percent in the price shocks, resulting in a standard deviation of 16 percent. The low risk case leads to a standard deviation of 4 percent in the random walk and of 6 percent in the price shocks. The volatility of coal prices is assumed about two-thirds that of gas prices in both cases. Load Growth. Unexpected changes in loads can affect the utility's revenues, profits and customer prices. TU's 1995 load forecast predicts an average rate of growth in peak loads of 2.5 percent (compared with the historic 3.4 percent) from 1994 to 2004, with a 40 percent chance the rate may reach 3.9 percent and a 40 percent chance that it may fall to 0.9 percent. Loads are modeled the same as fuel prices, with a combination of a random walk and one-year load shocks due to weather and other factors. Based on TU's projections (and taking into account planned demand-side management efforts), we assume a mean rate of increase of 1.93 percent, with a standard deviation of 3.8 percent to match the range of TU's 10-year forecast. One-year load shocks are allowed a standard deviation of 3.25 percent, for total standard deviation in load changes of 5 percent. Environmental Costs. Environmental regulatory risks are more difficult to simulate because limited historical data exists. The potential liability for electric utilities and their customers appears large. According to Energy Information Administration data, investor-owned utilities have invested about $60 billion in environmental compliance costs over the past several decades; TU electric's cumulative investment is $2.4 billion. The greatest future cost may come from greenhouse-gas regulation. For simplicity, all potential environmental regulatory costs are represented as a CO2 tax or fee. In the high-risk case, the probability is 70 percent over the 20 years after the first year of operation of the wind or gas plant that such regulation will occur. In the low risk case, the probability over 20 years is 30 percent. We believe a fair range of estimates for the probable cost of control would be $5 to $35 per ton, with a mean value of $25 per ton. Values are drawn from the right half of a normal distribution with zero mean and $31.3/ton standard deviation. Wind Plant Availability. Uncertainty in plant availability is often ignored in utility resource planning, although it can have a powerful impact on reliability and cost of service. It is especially important in this study because the availability of wind plants is likely to vary more than fossil-fuel plants. The variability in the annual output of wind power plants is well understood and easily modeled. To estimate its magnitude we simulated the performance of a wind plant using the Enercon E-40 wind turbine and four years of wind data collected in the DOE Candidate Wind Site program near Amarillo, Texas. The resulting annual average capacity factor of the wind plant is approximately 36 percent (assuming a 5 percent wind speed reduction due to wake losses and a 2 percent average power reduction caused by individual turbine outages), with a standard deviation of 6.5 percent. The uncertainty in wind plant output is incorporated into the model by randomly selecting a capacity factor in each year from a normal distribution with the given mean and standard deviation. When the capacity factor is lower than expected, the model draws more generation than usual from fossil resources. When the capacity factor is higher than expected, the opposite occurs. Estimates of fluctuations in the availability of fossil-fuel and nuclear plants were derived from five-year historical data for numerous plants published in the National Electric Reliability Council's Generating Availability Report. Since the figures in this report no doubt include some plants that are especially prone to failure, we scaled down the resulting estimates for this study. For existing plants and new coal plants, we assume a standard deviation in FOR of 10 percent. For gas-fired combustion turbines and combined cycle units, standard deviation was fixed at 5 percent.
Articles found on this page are available to Internet subscribers only. For more information about obtaining a username and password, please call our Customer Service Department at 1-800-368-5001.