Energy Risk & Markets
How to price new load-servicing contracts while incorporating market-risk analysis into such deals.
Why have basic generation service auctions historically been overly competitive given the prevailing market prices at the time? The answer requires an exploration of the concept of "charging" for market risk and then incorporating the existing risk profile of the bidding organization. EnergyCo- a hypothetical yet typical 5,000-MW vertically integrated energy company with a relatively balanced portfolio of generation, customer load, and wholesale trades-will help illustrate these points, showing how an organization can price a new load-servicing contract in isolation and then in conjunction with its existing portfolio.
Generation and customer demand have complicated dynamics. They are not easily understood because they are subject to a number of simultaneous uncertainties. The market-price uncertainty related to power and fuel prices create a spread-option portfolio that could be managed using financial derivative theory. However, generation plants also are subject to a range of operational factors (, start-up and shut-down costs, ramp rates, forced outages, environmental constraints, etc.) that have a material impact on the plant's performance. Customer demand has a significant level of volumetric uncertainty as well. Much of it can be explained by the region's weather conditions.
A merchant must consider these factors to understand the risk a new trade brings to an existing business. Monte Carlo simulation is one effective way of modeling all of these factors simultaneously. Simulation allows for prices and temperatures to be simulated and fed into generation dispatch and customer load models. In this way, we can generate 1,000 or more scenarios of hourly forecasts of generation, demand, mark-to-market transactions, and settlement.
This article uses Monte Carlo simulation to generate scenarios of net earnings for EnergyCo's current portfolio and a potential new load-serving contract. This information is then used to generate values and risk measures and risk-adjusted return on capital (RAROC).
EnergyCo serves a range of residential, commercial, and industrial customers. It forecasts its production and supply needs using Monte Carlo simulation modeling that incorporates the uncertainty in power and fuel prices, regional temperatures, and customer-load behavior to understand its expected net energy position. This analysis forecasts an expected excess level of supply of 1.5 million MWh during the next two years. Figures 1 and 2 summarize EnergyCo's net position.
At first glance, this portfolio appears to be relatively balanced between generation and load when the trading portfolio is taken into consideration. However, the drivers of this balance (, market prices, expected forced outage rate, temperature, customer demand behavior, etc.) create significant uncertainty in the actual balance. A significant excess or deficit could lead to enormous volatility in earnings and cash flows. When this balance is evaluated at the monthly level, the uncertainty is even more pronounced. For instance, Figure 3 illustrates the range of monthly imbalance that is expected for EnergyCo. While Apr. '06 and Sept. '06 are two examples of a balanced month, there is still a significant level of risk that the balance could be off by more than 500,000 MWh. Many of the other months have potential imbalances up to twice that level.
Extending the Forecast
This analysis can be extended to generating a comprehensive set of correlated net earnings forecasts. For example, EnergyCo can take the volumetric forecasts discussed above, plus a set of corresponding power and fuel price simulations and related operating costs to arrive at the earnings forecast. Figure 4 illustrates an example of a set of simulated spot-power prices during July 2005.
This analysis allows EnergyCo to forecast a net operating income of $1.2 billion ($659 million in '05, $555 million in '06) during the next two years. They also have estimated $203 million in earnings at risk (expected earnings less the 5 percent worst-case earnings). Figure 5 summarizes the results of the simulation analysis.
EnergyCo now understands its expected earnings and risk. These measures can be combined to generate a RAROC. The benefit of this is that RAROC allows for risk/return comparisons across different business lines. The enterprise-wide earnings forecasts summarized above can be generated for individual business activities. If we isolate these business lines we can generate a RAROC for each. Figure 6 summarizes the expected earnings, worst-case earnings, EaR, and RAROC for EnergyCo's three activities. When comparing each activity, it is clear that generation is the most attractive because it has the highest RAROC (1.97). However, there is another benefit to these other activities. They are risk reducing when aggregated with generation. This combination improves the RAROC of generation from 1.97 to 5.97 for the total enterprise.
EnergyCo is now armed with all of the tools it requires to price new trades or load-servicing commitments. Given its simulated forecasts for its current portfolio of exposures, it is in a much better position to understand its needs for additional trades. By layering on the simulation results of a contemplated trade, the organization can review the trade's overall impact on its net position.
For the purposes of illustration, suppose EnergyCo has the opportunity to enter into a load-serving contract for 115 MW of additional load (0.924 million MWhs over two years). However, it is a competitive market subject to bids from multiple merchants. EnergyCo must make a competitive bid that is low enough to compete with the market.
What should the bid price be?
Pricing is complicated. The merchants needs to incorporate the market price as well as a range of pass-through costs. For example, any merchant bidding on the BGS auction must include the following costs in their bid price:
Transmission losses; Capacity and FERC charges; Green costs; Operating reserves - real-time and day-ahead; Reactive supply and voltage control; Regulation and frequency response; Transitional market expansion; Transmission charges; and Transmission owner schedule/control dispatch service.
In addition, there is a price risk and volumetric risk that should be incorporated into the bid.
Step 1: Determine the Bid Price to Generate a $0 NPV.
A rational first step in determining the bid price would be to pinpoint the rate at which the deal would generate a flat net present value (NPV) of $0. This is the price you would have to collect per megawatt-hour to cover all the costs included in the commitment. The uncertainty in prices and customer demand likely would make the results higher or lower than this, but on average, the merchant could expect a balance between revenues and costs.
A Monte Carlo simulation model can be used to generate a distribution of load and price scenarios. Based on the analysis, a bid price of $52.96/MWh is determined to be the price at which this deal will expected to return a $0 NPV. Figure 7 summarizes the risk and distribution of outcomes for this contract.
Although the NPV is $0 under this bid price, plenty of scenarios would result in a loss for this deal. In fact, there is a 5 percent chance that the loss will be greater than $8.56 million. The absolute worst scenario results in a $12.78 million loss. One would argue that the possibility of losses in this range requires the bidder to add a premium onto their bid to generate a positive expected NPV.
Step 2: Incorporate a Risk Charge into the Bid Price.
Before a risk charge can be determined, we must define the concept. The question to ask is, "What return must be realized in order to compensate the bidder for the risk they are taking?" The "risk charge" is simply a built in premium to the bid price to compensate for the cost of risk. One common approach is to take the total risk of the trade, in this case the Earnings at Risk (EaR), and treat it like investment capital.
The level of return on risk capital depends on a number of factors including risk appetite, corporate strategy, alternative uses of this capital, etc. For illustrative purposes, assume EnergyCo requires a 100 percent return on every incremental dollar put at risk. In this case they would require a return of $8.6 million. Therefore, the bid price must be set at $62.79/MWh (almost $10 above the zero NPV bid price). At this price, the expected return would be sufficient to cover the required risk charge.
Step 3: Determine Bid Price Based on Incremental Risk Impact on Current Portfolio.
Recent load-serving auctions have been very competitive. It is likely that such a premium would not be accepted in a competitive market. However, any merchant with an existing portfolio could still benefit from a lower bid price if they consider how this transaction would impact the risk profile of their pre-existing portfolio.
If you add the transaction to EnergyCo's current portfolio, it reduces the overall portfolio's earnings at risk by $738,000. One can view that as releasing $738 of risk capital. As such, it is now possible to bid a price that actually returns a value less than zero. In this example, EnergyCo can now offer $52.30/MWh and maintain its current return-over-risk ratio. In this case, the expected net present value is estimated to be a loss of $738,000.
Step 4: Use Portfolio RAROC for Setting Bid Price.
While the overall portfolio's risk in the previous price has been reduced by $738,000, the RAROC increases from 5.66 to 5.68. In other words, by executing this trade at an expected loss, EnergyCo has actually maintained its NPV and improved its risk/return profile.
If the organization wants to be even more competitive in its bid for this deal, it could offer even lower rates and maintain its risk/return profile. It could generate a loss of $4.2 million on this trade and maintain the original RAROC of 5.66. This would allow the bid price to be as low as $48.19/MWh and still maintain the 5.66 RAROC. This gives the bidder enormous flexibility in acquiring more business.
The analysis presented in this article illustrates the volumetric and price risks that should be incorporated in the potential bid price for load servicing contracts. We use Monte Carlo simulation to estimate a range of bid prices as low as $48.19/MWh and as high as $62.79/MWh. This highlights that risk and the existing portfolio of a bidding organization have a material impact on the determination of the proper bid price for any transaction, especially load servicing.
Articles found on this page are available to subscribers only. For more information about obtaining a username and password, please call our Customer Service Department at 1-800-368-5001.