As this article goes to press, the news is full of stories about Calpine and the difficulty merchant generation players face from the uncertainty and volatility of power markets.
With that in mind, now is a good time to review key measures of performance and profitability under uncertain conditions.
One method of determining the gross margin (total revenue minus cost of inputs) available to generators involves using electricity and natural-gas prices to derive the market heat rate. This is calculated by dividing the forecast electricity price by the forecast natural-gas price at the burner-tip.1 This normalizes the electricity price forecast relative to the natural-gas price forecast and allows one to observe trends in the long-term electricity price forecast that are independent of changes in natural-gas prices.
An increasing market heat rate corresponds to an increasing gross margin for gas-fired electricity producers. A rising market heat rate provides some information on the actual heat rate of the unit on the margin, but also includes the price impact of generators bidding higher than variable costs during periods of relative supply scarcity.
Figure 1 presents the market heat rates implicit in Global Energy’s current electricity price forecast for Entergy, based on monthly on-peak and off-peak electricity prices. The monthly pattern in market heat rates illustrates the uplift in summer prices, which is due to the higher percentage of time that electricity is generated using natural gas, and the generally higher (less efficient) heat rates of the equipment setting these prices. This also reflects the premiums generators are able to capture during periods of relative scarcity. These market heat rates increase during the first 10 years because of a declining reserve margin as the overbuild subsides and the growing percentage of time that natural gas is marginal.
For those interested in natural gas-fired generation, it is customary to describe an energy price forecast in terms of a “spark spread.” Spark spread is an indication of the profit per megawatt-hour that a gas-fired generator might expect under a particular energy price forecast. This term (spark spread) is generally used for new, highly efficient natural gas-fired generation, such as combined-cycle gas-fired capacity, and in relationship to heavy load (peak load) electricity prices, rather than all hours of the year.
These new, highly efficient gas-fired generators will use approximately 7,000 Btu of natural gas to generate 1 kWh of electricity. Therefore, spark spreads often are quoted in the trade press as being the spot-market price (during heavy load hours for a month) for 1 kWh of electricity less the cost of natural gas incurred in a 7,000 Btu/kWh generator.
In the Southeast, for example, natural gas-fired generation is “on the margin” frequently. This affects spot-market electricity prices. While a high electricity price is always good if you are selling output of a wind project, it is not necessarily good (or as good) if you have a natural gas-fired generator. If natural gas prices are high, then the marginal natural gas-fired generator will not experience improved profitability from the commensurately higher spot-market electricity prices.
Global Energy forecasts spark spreads for all market areas. Figure 2 shows forecast monthly spark spreads for Entergy under Global Energy’s Fall 2005 forecast. For comparison, we also show spark spreads based on a 10,000 Btu/kWh heat rate.
Higher natural-gas prices generally result in higher spark spreads. This rule of thumb derives from the fact that in heavy load hours generators with lower efficiency in converting natural gas to electricity are dispatched. For example, in heavy load hours a simple-cycle gas turbine with a heat rate of 10,000 Btu/kWh could be called to meet load. As a result, spot-market prices during these hours will need to be high enough to cover the operating cost of the 10,000 Btu/kWh unit. So, if gas prices increase by $1/MMBTU, then spot-market prices during heavy load hours should increase by $10/MWh to cover the increased cost of the 10,000 Btu/kWh unit. However, the cost for a more efficient (7,000 Btu/kWh) combined- cycle gas plant only will increase by $7/MWh. Therefore, the spark spread for the 7,000 Btu/kWh heat rate combined-cycle unit will increase by $3/MWh under this scenario.
Spark spreads indicated by traders occasionally are higher than those developed from fundamentals-based energy price forecasts. Trader data allegedly is based on actual deals being made in power markets. Global Energy has attempted to verify the existence of actual trades behind trader-quoted numbers, but it often is difficult to get information that can validate that actual deals being made at these high spark-spread levels. It is possible that some small quantities are being bought and sold at these levels. However, there often does not appear to be a significant amount of volume in power transactions to make for reliable decision making.
It also is possible that long-dated trader spark-spread data are the mid-level of a “buy-sell” spread. The “buy-sell” spread is likely to be large in these outer years, and the asking (sell) price may be at a level to cover the full cost of a new generator (fixed and variable). If “asking” prices are at this level, the mid-level of a “buy-sell” spread may be artificially high. We believe the most reliable forecast is a fundamentals-based forecast assuming normal conditions, which might differ from what certain buyers and sellers are willing to trade at as a result of their respective concerns about future uncertainties.
Any number of factors can lead to differences between fundamental forecasts and forward prices. In particular, there may not be a liquid market for forward contracts between creditworthy parties. Where illiquidity exists, forward prices can be biased by relatively small-value transactions. Forward markets also reflect a perceived “risk premium” that is absent from our deterministic spot-price forecast. If prices are perceived to increase sharply under abnormal conditions (e.g., higher expected load), a premium may be built into the forward price to reflect this. Finally, differences in perceived weather conditions will result in different projections. Both the fundamentals-based forecast and all forward markets are “forecast” or projected using potentially differing input assumptions.
While we know that traders prefer futures as a benchmark for deals, we believe that fundamentals-based forecasts provide a quantifiable basis for determining the resources likely to be on the margin and hence a reasonable indication of future spot-market prices. Where differences are identified between the fundamental forecast and forward markets, market participants would best be served by understanding the basis for both forecasts.
Figure 3 compares historic with our forecasted wholesale on-peak prices through 2008.
Power prices have remained fairly volatile due to high natural-gas price volatility. We are witnessing the longest period of high gas prices in recent history, far surpassing the temporary spikes observed during most of the 1990s and early parts of this decade. Power prices are forecast to fall in the near term as a result of falling natural gas prices in combination with high reserve margins across the Southeast.
Figure 4 compares the forecast with actual on-peak market heat rates. As the Southeast markets gradually grow out the historic overbuild, we expect market areas within the region to witness some heat-rate recovery in 2006 and beyond. By 2008, summer on-peak market heat rates will approach 12,000 BTU/kWh. Thereafter, market heat rates in the Southeast will continue to rise, signaling gradual market recovery.
Volatility, due to many factors, has the potential of moving a market like the Southeast into the higher-cost portion of the supply curve where scarcity premiums can come into play. Participants in power markets in the Southeast need to be aware of these possibilities and have contingency plans for dealing with them.
The traditional ways of evaluating power generation assets can be seen to have the following shortcomings:
Starting with a consistent price forecast, developing the stochastic parameters and then running the alternative simulated price paths through a full dispatch model often is the most appropriate methodology for generation asset valuation under conditions of uncertainty or volatility.
Generally, the term “stochastic” is used to indicate that a particular subject is seen from a point of view of randomness, as part of a probability theory that can predict how likely a particular outcome is.
An important issue is the granularity of the starting price models. In this case we start with the hourly power prices that have been directly linked to the daily gas prices. This allows us to disaggregate volatility and correlations down to the daily level (the minimum gas-price period) and ensure these critical profiles are not lost by an averaging process. More important, we are able to project the changing relationship between gas and power prices through time. There are thus two “random” factors that affect electricity and fuel prices.
Long-run (LR) factors such as technology, population changes, and gross domestic product (GDP) differences will result in a long-run random effect on prices. Long-term volatility tends to be small compared to the short-term shocks. These random effects will have a limited effect on individual years particularly in the near term, but an increasingly important effect over the long term. The effect will be to show an increasing variance over time. We assume that LR volatility does not mean revert and follows a standard Brownian motion process.
Random factors such as weather, outages, and short-run liquidity effects will be captured in the short-run volatility parameter. These short-run “shocks” are assumed to be temporary deviations from the equilibrium. This process tends to be more significant in driving what is commonly perceived as price volatility and will capture the now infamous price spikes within the electricity price process.
Global Energy and others throughout the industry have concluded that the short-run shocks are mean reverting. In other words, they will revert to the equilibrium price after some time has passed.
The mean-reverting process can be likened to applying a piece of elastic between the observed price and the equilibrium price. A random factor continues to be applied to the price as it moves through time, but as it moves further away from its equilibrium price a proportionately increasing force is applied to it to pull it back. The speed of mean reversion, a key input variable in this process, determines how quickly prices revert to equilibrium.
Once we have identified the short-run and long-run parameters, it is necessary to calculate the related correlations. In this analysis we are correlating all the fuel and electricity prices within a region for both the short- and long-run conditions.
Using our Planning & Risk stochastic simulation tool, Global Energy developed a range of prices that could occur in the future. Figure 6 shows the expected average monthly electricity price in Entergy compared to the 10th and 90th percentiles of the price forecast. The stochastic results do not represent a true confidence band that spans multiple months. Rather, these results reflect forecast “error bands” indicating that the resultant price distribution for each month is independent of the stochastic behavior depicted in previous months.2
As part of its ongoing analysis of regional power markets, Global Energy examined the performance of three existing generation technologies under our current Global Energy Reference Case Price Forecast:
Each technology was assessed using deterministic and stochastic market prices.
We examined a typical steam unit with a 9,800 Btu/kWh full load heat rate and average operational parameters. The combustion turbine was modeled using a 10,500 Btu/kWh heat rate. Finally, the combined-cycle unit had a full load heat rate of 7,100 Btu/kWh.
In this assessment, the unit value was examined using two approaches. In the first, we examined value in a deterministic setting in which fundamentals based fluctuations in the expected load, gas, and hydro generation were not captured. Global Energy defines the resulting net operating revenues under this setting as deterministic or intrinsic value. Under the second approach, we assessed value in a setting in which volatility is captured through a set of estimated parameters that represent the inherent uncertainty of load, gas price, and hydro generation in the Southeast region.
In this stochastic analysis, the inclusion of volatility allows one to capture the optionality value of these generating units. A unit with greater flexibility to respond to volatile fuel and fundamentals-driven market clearing prices can capture additional revenues beyond those shown in the deterministic analysis alone. This is often referred to as the plant’s real option value. The difference between this value and the deterministic value is called a stochastic or extrinsic value associated with volatility.
To calculate deterministic and stochastic value, we used our Planning & Risk model, which explicitly permits the input of spot electric price volatility and gas price volatility and seasonal correlation parameters. Historical data in the Southeast region was used to derive these parameter estimates. Two-hundred-and-fifty Monte Carlo estimates of future market outcomes were projected to estimate the plant’s real option value.
Overall, stochastic analyses capture the optionality value of a flexible resource missed by deterministic analyses. However, advanced trading schemes such as complete delta hedging are necessary to capture the full real option value indicated in our analysis. Of course, this also requires significantly greater trading liquidity for future prices than what presently exists in today’s power markets.
The asymmetric distribution of market-clearing prices produces a greater probability that prices will increase substantially under adverse circumstances than they will decrease under favorable circumstances. This is especially true in a geographic area such as the Southeast region where natural gas plays a dominant role in the power market with respect to both energy supply and reliability.
Volatility and uncertainty are permanent fixtures in every regional energy market. Understanding how some key measures of margins, profitability, and performance under uncertainty can affect asset valuation, materially can affect the performance of an asset or portfolio.
1. The formula is Market Heat Rate = Electricity Price ($/MWh) / Fuel Price ($/MMBTU) x 1,000.The multiplication by 1,000 is to correct for the net impact of converting the dimensions of the equation. In trading terms, a generator with a heat rate lower than the market heat rate is “in the money” and can profitably sell electricity.
2. The percentile prices, sometimes referred to as P10 and P90 prices, indicate an 80 percent confidence band for the price forecast. In other words, the 90th percentile indicates that in any given month, there is a 90 percent chance that prices will be at or below this price. Similarly, the 10th percentile indicates a 90 percent chance that prices will be at or above this level.