Retail Risk-Based Pricing
A new approach to rate design.
March 2004
By Gary Dorris and Sean Burrows
As energy markets have evolved in the late 1990s away from cost-based transactions to competitive market-based transactions, the exposure to market risks for the variable cost of supply has substantially increased.1 Reflected in these market risks are the diminishing reserves for North American gas supply, which has created conditions of extreme volatility in gas supply. The added market risk is compounded by the sensitivity of some retail load customers to weather conditions. Through integrating load and price uncertainty, risk-based pricing (RBP) determines the true cost of service and provides a more comprehensive approach to developing retail rates in today's energy markets.2
Traditionally, retail pricing for regulated entities has been based on the average cost of service, and no factors are associated with the relative volumetric risks and the relationship to prices.3 The principal concept behind RBP is that customers with greater risk require greater working capital set-asides to address anomalous or unexpected events. RBP provides a structured approach to value the elements of retail supply cost risk and provide a systematic structure to incorporate these risks into retail prices.4 As a company's stock price carries a beta value with respect to movement in the market average, RBP introduces a method for reflecting the joint relationship of volumetric and price risk that effectively incorporates each customer's beta into retail electric rates. For competitive retail offerings, RBP provides a well structured approach to identify low-risk customers and develop a consistent risk metric for retail offerings.
Utilities are pressed with the challenge of providing service at the forecast cost of supply under highly volatile market and weather conditions that can leave utilities short of the cash required to maintain operations. Over the last three years, utilities have suffered a spate of rating agency downgrades, disabling disallowances related to fuel and purchased power, and, in a few instances, bankruptcy. In most cases, the financial hardship pressed upon utilities has been related to cash flows and the ability to cover current and future liabilities. RBP serves as a defensive mechanism that updates previous rate-making strategies for consistency with current market structures. RBP is a natural extension of real-time pricing initiatives that works contemporaneously with time-of-use pricing and multi-tiered tariffs.5
Volumetric Risk
Understanding the variability in weather is important for accurate characterization of electricity load, price volatility, and volumetric risk. By producing accurate simulations of spatial and temporal weather patterns, we are able to assess the volumetric risk and weather risk of different retail customers.
For example, looking at Figures 1a and 1b, we have two different customer class profiles in the ERCOT Coastal Zone for July 2004. The first profile, ERCOT "Business High Load Factor" (BusHiLF), is a commercial profile where the daily energy uncertainty is around 16 percent (see Figure 1a). The second profile customer, ERCOT "Residential Low" (ResLow), is a residential profile where the daily energy uncertainty is 26 percent (see Figure 1b). A summary of the annual weather-based uncertainties for the different ERCOT zones and profiles clearly shows large differences between the different customer profile types, with a minimum of 14 percent for "BusHiLF" and a maximum of 53 percent for "Residential High" (ResHi) (see Table 1).
Since the daily energy uncertainty is caused by variability in weather, and prices are correlated to weather through system energy demand, we have observed that increased volumetric uncertainty can lead to increased cost and risk of service. We can use the differences in volumetric load uncertainty of different customers to develop risk-based prices that account for the observed uncertainty.
Cash Flow at Risk
For most utilities, the principal element of uncertainty in cash flows is related either to the cost of supply or the quantity demanded. The most serious issues that affect the future financial viability of a utility are unexpected costs and inadequate revenues. The impact of customer demand and price on a utility follows from the cash flow equation of gross margin.6 The only variable in the gross margin equation that does not have a random component is the customer price Pni. Thus, our problem becomes focused on the optimal Pni to reflect the market and volumetric risk associated with serving customer load Qni. Since customer load and cost of service are stochastic variables, we need to jointly simulate the components of the gross margin function to assess the individual elements of risk posed by a customer. For example, a fuel oil pumping station may have highly volatile demand that is uncorrelated with electricity prices, whereas a commercial office building has demand closely correlated with prices. Through simulation, we are able to capture the nature and magnitude of uncertainty surrounding each stochastic variables and their joint relationship. The result is a distribution of gross margin for each customer.
 |
Risk can be systematically measured through examining the difference between the mean and the fifth percentile of the gross margin distribution.7 We refer to this difference as the gross margin at risk (GMaR), as illustrated in Figure 2 on p. 47.
The mean gross margin (green dashed line) is $2.9M, and the 5th percentile (red dotted line) is -$2.4M, so the GMaR (solid black line) is $5.3M.
RBP seeks to reduce the difference between the mean and the lower tail of the gross margin distribution for the entire utility by selectively setting prices higher for customers with greater risk and reducing costs for customers that pose less risk. RBP prices can be scaled to achieve the same expected revenues for the utility, but they afford some additional downside protection against extreme events that affect the cost of supply.
Most economists would insist that assessment of the marginal contribution of risk should be determined relative to marginal costs or opportunity costs of the spot market.8 Under the opportunity cost criteria, the gross margin equation would simplify with the elimination of the fuel and variable cost components, leaving the cost-of-service analysis exclusive with respect to the market. However, a utility's own cost of supply combined with the costs or earnings through daily balancing exchanges may be a more fitting first step for most regulated entities. These results can then be compared with assessment of risks relative to opportunity costs of the market.
RBP Development
RBP seeks to allocate the cost of energy supply to reflect the relative risk of each customer. For traditional utilities, the variable costs of production are commonly carried across all customers on an average-cost basis that remains independent of time of use. An appropriate incremental step toward full RBP is to begin allocating the variable cost of production to appropriately reflect the risk in supply costs. This leaves the additional opportunity of allocating fixed costs with a risk component. For competitive retail providers, the RBP becomes a strategic necessity and can be a significant differentiating factor in long-term success.
The RBP formula consists of the original retail price developed from the expected value of the cost of service, plus (positive or negative) a risk adjustment factor, plus a term to adjust rates up or down to equal an enterprise gross margin revenue requirement.9 The risk adjustment factor corresponds to the second component in Equation 2 and increases or decreases rates from the original price. The third component of Equation 2 increases or decreases rates to achieve a target gross margin revenue requirement. This last term would not be applicable for competitive retail offerings because of the lack of revenue requirements in deregulated markets. Before demonstrating the impact of RBP on retail rates and gross margin revenue, the analytical requirements to determine RBP need to be defined further.
Analytic Requirements for RBP
To generate accurate portrayal of the uncertainty in gross margin by customer, we apply an integrated simulation framework. Our framework captures the range in outcomes for each element of the gross margin function along with their covariate relationships. Through a series of linked structural state-space models, the fundamental uncertainties between weather, load, and market prices (and generation costs) can be captured. By capturing this fundamental relationship, the analytical foundation necessary to develop RBP can be developed.
An overview of the simulation engine we have developed to support RBP is presented in Figure 3.10 The data inputs consist of traditional market fundamentals of demand and supply accounted through variables of system and customer load, supply stack, transmission constraints, fuel prices, forward curves, and reserve margins. The input data flows into the simulation engine to model uncertain electricity market drivers, supply resources, and market prices. The uncertainty in each of the structural variables is represented by the middle row of boxes in the simulation engine. The output of the simulation engine produces realizations of cash flows that are aggregated into distributions of cash flows for gross margin cost of supply and customer load.
By first simulating weather at multiple weather stations across the region, we maintain a key relationship of weather driving regional load and weather driving customer load. Thus, we preserve the temporal structure of a customer's load response to weather and regional load. This relationship of weather and load translates into the cost of service through the effect of regional load as an explanatory factor of market electricity prices. Figure 4 illustrates the relationship of system load on prices.11
The uncertainty of electricity prices and the variables that drive prices are simulated using a structural state-space model with regime switching. The model captures the structural elements that drive electricity prices and the stochastic elements of uncertainty around these fundamental drivers, plus the unexplained stochastic noise in electricity prices.
Applying RPB to a Retail Portfolio
In this section, we present an example of applying RBP to a small retail portfolio. Our example examines the effect of applying the RBP methodology to two retail portfolios in the ERCOT market: 1) a group of BusHiLF load profile customers; and 2) a group of large commercial office buildings.12 The portfolio of BusHiLF customers has relatively low weather sensitivity and volumetric risk compared with other load profile customers, as shown in Table 1. The portfolio of commercial office buildings (Commercial) has nearly the same absolute volumetric risk as the BusHiLF portfolio but carries a larger sensitivity to changes in weather as opposed to changes in office occupancy rates. By having two portfolios with comparable volumetric risks but different weather sensitivities, we further illustrate the importance of accurately articulating the weather, load, and price-risk relationship as part of RBP and of reducing the amount of gross margin at risk.
The traditional pricing method for energy13 has been developed on a expected value based on the average cost of service plus a fixed margin.14 The BusHi portfolio has an original wholesale rate of $44/MWh compared with $43/MWh of the Commercial portfolio. By applying the RBP formula in equation 2, the average retail prices change so that the BusHi portfolio is $41.7/MWh and the Commercial portfolio is $45.3/MWh. The BusHi portfolio's RBP rate decreases by $2.3/MWh because load has a relatively limited sensitivity to changes in weather (i.e., a small weather sensitivity). The Commercial portfolio's rates increase because its load is highly correlated with weather and weather has a pronounced correlation with price in ERCOT.
The changes in pricing from traditional rate design methodologies to the RBP rates significantly reduce the uncertainty in gross margin at risk. The expected gross margins shown in Figure 5 over a 12-month cycle, corresponding to the upper lines for the combined portfolio for BusHi and Commerical, are $2.5 million under both pricing structures. Although the portfolios have the same expected gross margin over the annual forecast horizon, the RBP line contains significantly smaller fluctuations in month-to-month gross margin revenue. RBP has a pronounced effect on reduced volatility in gross margin at the 5 percent level. The annual gross margin at risk is $3.8 million under the original rate structure versus $2.9 million under RBP rates. Thus, RBP achieves a 30 percent reduction in GMaR over traditional rates while maintaining the same expected revenue.
The benefits of RBP are pronounced, but these benefits are achieved only through application of the appropriate analytical rigor. RBP requires a complex series of integrated simulation models to capture accurately the nature and magnitude of each customer's cost of service risk. It is easy to muddle volumetric risk with weather sensitivity. However, we have shown that accurately delineating the nature of volumetric risk derived from weather results in different retail prices than under more traditional approaches.
RBP improves a utility's bottom line through reducing the effect of more weather-sensitive customers on cash flows. The result of RBP is reduced cash flow variance from extreme weather events and market events. By more effectively pricing risk of each customer class' cost of supply, RBP reduces the volatility of cash flows. In turn, this reduces the cash reserve requirements. This freed capital can find more product uses with the applied RBP as a mechanism for helping preserve capital adequacy.
Beyond positively affecting a utility's bottom line, RBP heeds to pricing rationalism for both regulated entities and competitive retail offerings. By introducing individual customer characteristics and associated cost-of-service risks into a unified pricing methodology, retail electric prices more closely approximate the marginal cost of production, and the extent of cross subsidies can be better known. In economic terms, inclusion of the cost-of-service components creates additional pricing efficiencies that, on average, benefit consumers. The gains for both customers and utilities make RBP an attractive option to develop electric rates for today's energy markets.
Endnotes
- Nationally, energy markets are starting to evolve toward a common structure with regional transmission organizations (RTOs) and locational market prices (LMP).
- The true cost of service follows conditions where the marginal cost equals the marginal revenue.
- Bauer, John, American Economic Review; March 29 Supplement, Vol. 19, Issue 1, p. 219.
- Commonly, utilities calculate the cost of fuel and purchased power equally among all customer classes on an average-cost basis.
- Hartman, Raymond S, Jensen, Kenneth A, Seiden, Kenneth P., Applied Economics. London: Apr 1994. Vol. 26, Iss. 4; p. 363. Gross Margin as defined here is equivalent to EBITDA.
- Dorris, Gary; Dunn, Andy, Electric Light & Power, October 2001, Vol. 79, Issue 10, p. 32; (AN 5423572) "Making the Shift to Earnings at Risk."
- Hotelling, Harold. "The General Welfare in Relation to Problems of Taxation and of Railway and Utility Rates." Econometrica July 1938, pp. 242-69.
- Companies developing competitive retail offerings typically do not have an enterprise gross margin revenue requirement, leaving the third term as zero.
- The commercial software application used to model energy load and markets and develop RBP is PowerSimm.TM
- Structural state-space models reflect uncertainty of the effect of each explanatory variable on price as well as the unexplained component. The explained component contains uncertainty in the parameter estimate of each fundamental variable influencing price. The unexplained component captures the random noise in electricity prices that cannot be explained by fundamental variables or time series terms. The split regression shown in Figure 5 contains a relatively modest amount of noise in electricity price of +/- $5/MWh when load is less than 44 GW. When load is greater than 44 GW, the unexplained noise "switches" to a higher state and captures uncertainty in prices of +/- 40/MWh.
- The influence of each structural variable on market prices follows from maximum likelihood regression analysis.
- These are interval data read (IDR) customers.
- Exclusive of T&D charges.
- Because the two customers have the same absolute volumetric risks, the fixed percentage adder is equivalent to the expected cost of service plus a non-weather based volumetric risk component.
Gary Dorris and Sean Burrows work for Ascend Analytics in Colorado. Contact Dorris at gdorris@ascendanalytics.com and Burrows at sburrows@ascendanalytics.com, or 303-415-1400.
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.
Public
Utilities Fortnightly & Spark
