The consequences of short-sighted rate making.
Natural gas local distribution companies (LDCs) charge customers for delivery service through what are commonly referred to as base rates.1 An LDC's base rates are set by its state regulatory commission in a general rate case and are intended to provide the LDC with a reasonable opportunity to recover its revenue requirement, or its operating expenses (including depreciation and taxes) and a reasonable return on the capital invested to provide gas distribution service.
Commissions set LDC base rates for each customer class. These typically contain a fixed monthly charge, generally called a customer charge or a facility charge, and a volumetric charge for each unit of gas delivered.2 The portions of the delivery revenue stream derived from customer charges and from volumetric charges differ from one LDC to the next, but LDCs typically depend on volumetric charges for a substantial portion of their base revenue recovery. For example, several recent LDC rate-case filings show that between 46 and 57 percent of current residential delivery revenue is accounted for by volumetric charges, and 72 to 77 percent of current commercial delivery revenue is volumetrically dependent.3
Increasingly, many LDCs are discovering that traditional causes of inadequate earnings, such as higher wages and taxes and investment in new facilities, are not the primary drivers behind the need to request higher base rates. Rather than being driven by expenses or investment, inadequate LDC earnings are due, in large part, to delivery revenue shortfalls-specifically, their inability to achieve the delivery volumes used by their commissions to set volumetric base rates for residential and small commercial customers.4
Causes of Delivery Volume Shortfalls
There are two primary causes of LDC delivery volume shortfalls. The first cause is weather-related. LDC delivery volumes, especially residential and small commercial volumes, are very sensitive to winter weather conditions. In developing the volumetric component of LDC base rates, state commissions apply volumes that are expected to be delivered under normal weather conditions. If a commission uses an unrealistically cold measure of normal weather, the LDC will discover that it cannot consistently achieve the expected delivery volumes, and associated revenue and earnings shortfalls appear. State commissions throughout the country successfully have resolved this problem for a number of LDCs through adoption of weather normalization clauses.5 Through the operation of these clauses, LDCs can adjust the delivery service portion of customer bills to reflect volumes that would be delivered under the normal weather conditions used in setting base rates. Depending on the weather clause structure, the adjustments may be reflected on customer bills during the billing month in which the abnormal weather occurs, or they may be accumulated into a deferral account and reflected on next season's bills. Weather normalization clauses mitigate the problem of determining what normal weather is for the purpose of setting base rates, and also address the post rate-case earnings consequences of selecting an inappropriate measure of normal weather.6
The second cause of delivery volume shortfalls relates to a continuing pattern of declining residential and commercial usage per customer. Weather-normalized delivery volumes used is setting base rates are based on a 12-month period. Depending on regulatory practice in a state, this period may be a recent historical year, a partially forecasted year (e.g., six-months historical and six months forecasted), or a fully-forecasted year (, the first year when new rates are in effect). If usage per residential or commercial customer falls after the period used to set base rates, the LDC cannot achieve the expected delivery volumes, and earnings shortfalls result. With substantial dependence on volumetric revenue recovery, resulting earnings shortfalls can be significant, especially because it is not uncommon for an LDC to derive 90 percent or more of its delivery charge revenue from residential and small commercial customers whose usage is declining.7 Faced with these usage patterns, an LDC's only choice is to file new and frequent rate cases to update the volumes on which its rates are set.
While state commissions throughout the country have addressed weather-related causes of delivery volume shortfalls through adoption of weather normalization clauses, regulatory recognition of declining usage patterns is much less prevalent.8 LDCs must demonstrate the significance of these usage patterns on their earnings and provide their commissions with reasonable regulatory tools to address them. Properly constructed, regulatory responses to these usage patterns can be designed to benefit both LDCs and their customers, much like weather normalization clauses have been mutually beneficial.
The Significance of Declining Usage
The declining pattern of residential and commercial usage per customer is well documented. For example, in three studies, the American Gas Association (AGA) found that weather-normalized annual residential usage per customer in the United States steadily has declined from 1980 through 2001, with further declines projected through 2020, as summarized in Figure 1.9
These studies also demonstrate that the trend appears consistently over time in each of the four U. S. Census regions, with the exception of an increase in the Northeast between 1980 and 1990, as summarized in Figure 2.10
The pattern of declining usage differs by region, with the sharpest continuous relative declines appearing in the Midwest and South. Between 1980 and 2001, usage per residential customer fell by 24 percent in the Midwest and 22 percent in the South.
In addition to its residential studies, the AGA conducted a study of commercial sector gas usage. This study showed annual weather-normalized usage per commercial customer in the United States decreased by 18 percent (from 780 Mcf to 640 Mcf) over the two decade period from 1979 through 1999.11 Similar to its residential customer findings, AGA's regional analyses show particularly sharp commercial customer declines over this period of 27 percent and 30 percent in the Midwest and South, respectively.
The Regulatory Response
The analyses of company-specific residential and small commercial customer gas usage in the three case studies are consistent with findings of other studies in confirming the existence of continuing, declining residential and small commercial per customer usage over time. These patterns have important earnings consequences with traditional rate structures in effect today. Just as many LDCs and their state commissions effectively addressed weather-induced impacts on earnings through new regulatory mechanisms in the past, it is time for LDCs and their commissions to examine alternative mechanisms to address earnings impacts associated with these declining usage patterns.
Unless there are changes in the traditional rate-setting process, the only option for LDCs will be to continue to play catch-up through frequent general rate-case filings. While it is not my purpose to present and evaluate various regulatory tools to address the impact of these trends, I offer some observations that LDCs and state commissions should consider during the process to implement regulatory changes.
The fundamental issue relates to the reasonableness of volume-dependent base-rate structures. There are a number of sound regulatory policy reasons to move away from these structures. First, a base-rate structure containing significant usage-based charges is not supported by cost considerations. An LDC's cost of delivery service as reflected in its base-rate revenue requirement is driven almost exclusively by the number of customers served and the demands those customers place on the delivery system at peak times, not by the gas volumes that the LDC delivers to customers.12 Volumes that customers consume do affect gas-supply costs, but virtually all gas-supply costs typically are excluded from the determination of the delivery service revenue requirement and resulting base rates. Rather, gas-supply costs are recovered through separate cost-of-gas clauses. Accounting structures may result in some gas supply-related expenses being recorded as operating expenses includable in the delivery service revenue requirement, but these expenses are not a significant portion of LDC operating expenses. For example, load-dispatching expenses, which some analysts argue are volume driven, may be recorded as operating expenses.13 The fact remains that an LDC's delivery service revenue requirement is not significantly volume-driven, and cost considerations suggest application of a base rate structure that concentrates revenue recovery on non-volume-related rate elements.
Second, by breaking the link between an LDC's profitability and the volume of gas it delivers, an LDC will have a greater incentive to encourage its customers to use less gas. In advocating the need for regulatory authorities to consider proposals to decouple utility profitability from gas delivery levels, the AGA and Natural Resources Defense Council recently recognized the limitations inherent in volumetric-dependent rate designs:
When customers use less natural gas, utility profitability almost always suffers, because recovery of fixed costs is reduced in proportion to the reduction in sales. Thus, conservation may prevent the utility from recovering its authorized fixed costs and earning its state-allowed rate of return. In this important respect, traditional utility rate practices fail to align the interests of utility shareholders with those of utility customers and society as a whole.14
At times of high gas costs, well-designed conservation initiatives are particularly valuable to customers. By refocusing the manner in which delivery revenues are recovered, LDCs will have stronger incentives to propose these initiatives because their earnings would not be adversely affected and customers will benefit through savings in the cost of gas portion of their bills through reduced usage.
Third, the premise that low-use customers are low-income customers is not a sound basis for retaining a regulatory policy of recovering a substantial portion of an LDC's required delivery revenue from volumetric charges. Many low-income customers who live in old, poorly insulated structures use more gas than higher-income customers. And certain relatively low-use customers are middle- or high-income customers who, for example, heat their homes with electric heat pumps, using gas only for water heating and, perhaps, back-up central furnaces. As a result, a volumetric-driven rate design is a clumsy and ineffective vehicle to assist low-income customers. Some non-low-income customers unnecessarily will be assisted through the subsidy, and other deserving customers will go unaided. The low-income problem is more effectively addressed through special low-income rates, commission-approved weatherization programs, or expanded federal and state assistance grants rather than through rate design.
Fourth, concern about "rate shock" on low-use customers should not prevent a movement toward collection of a much greater portion of LDC required delivery-service revenue through fixed rather than variable charges.15 Rather, addressing this concern may require that the rate design changes be implemented gradually over time.16
Other regulatory mechanisms also can be implemented to address the impact of declining usage patterns. For example, a rate adjustment mechanism could be introduced that provides for periodic rate adjustments to true-up the dollar consequences of differences between customer usage actually experienced and usage levels for designing volumetric rates in the most recent rate case. Over time, if base-rate design changes ultimately are implemented that result in delivery revenue recovered largely through fixed-rate elements, such an adjustment mechanism could be phased out.
By effectively addressing declining residential and small commercial usage trends in an LDC rate case, the LDC and state commission can work together to improve the LDC's opportunity to achieve its authorized rate of return and the stability of its earnings and cash flow, encourage LDC involvement in conservation initiatives, and lower future customer bills through fewer rate cases, associated lower expenses and reduced gas usage resulting from conservation initiatives. Experience gained through implementation of weather normalization clauses suggests that regulatory solutions to LDC earnings problems can be beneficial to both LDCs and their customers. The same opportunity exists today in developing regulatory solutions to LDC earnings shortfalls resulting from declining residential and small commercial usage trends with traditional base rate structures.
- Those customers who secure their own gas supplies, directly or through the services of a gas marketer, pay the LDC only for delivery service. Customers who rely on the LDC for the commodity as well as delivery service also pay the LDC for the cost of gas that it incurs in securing supplies. LDCs typically recover gas-related costs through state commission-approved cost of gas clauses outside of general rate cases. While some LDC rate structures include a portion of gas costs in base rates, I use the term "base rates" as if these rates are designed to recover only LDC delivery costs. The discussion that follows remains applicable to LDC base rate structures that include recovery of some gas costs because, in these instances, gas costs intended to be recovered through base rates are trued-up through the operation of LDC cost of gas clauses.
- Customer class delineations differ among LDCs, but a typical class breakdown distinguishes among residential, small commercial (or small general service), large commercial, and industrial customer classes. Volumetric charges may be single per unit charges for all units delivered, or they may be stepped charges in which per unit rates vary with the level of consumption. The units of delivery for LDCs are either volumetric, i.e. hundred cubic feet or thousand cubic feet, or heat-based, i.e. therms. Some LDCs have rate structures that contain demand charges in addition to customer and volumetric charges, although these rate structures are typically applicable to large, non- residential customers.
- Missouri Gas Energy, Missouri Public Service Commission, ; Texas Gas Service Company, Railroad Commission of Texas, ; and CenterPoint Energy, Arkansas Public Service Commission, .
- Some LDCs face the additional problem of declining customer counts, a problem that exacerbates earnings inadequacies through continuing customer charge revenue shortfalls.
- Weather normalization clauses have been approved for various LDCs in states that include Alabama, Arkansas, Connecticut, Georgia, Kansas, Kentucky, Maryland, Mississippi, New Jersey, New York, Oklahoma, Oregon, Rhode Island, South Carolina, Tennessee, Texas, Utah, and Wyoming.
- While weather normalization clauses may eliminate rate case controversies concerning the definition of normal weather, a state commission should make every effort to select a measure that best reflects conditions that can be expected, on average, when rates are in the effect. This effort is needed to promote customer acceptance and understanding of weather clauses. For example, if a commission uses an unrealistically cold measure of normal weather in setting base rates, weather clause surcharges will appear on customer bills even during periods that most people would consider to be quite warm. In these warm periods, customers will complain about perceived overcharges resulting from the weather clause and will not understand how they benefit from the clause when, in fact, the problem lies in the base rate setting process.
- The use of forecasted periods for developing base rate delivery volumes may delay, but will not eliminate these earnings consequences.
- Two exceptions are worth noting. In California, Southwest Gas Corp. defers and recovers through rate adjustments the difference between recorded delivery charge revenues and rate case revenue levels [Core Fixed Cost Adjustment Mechanism, Cal. P.U.C. Sheet Nos. 6001-G and 6002-G]. In Oregon, Northwest Natural Gas Co. recovers 90 percent of lost residential and commercial delivery revenue through its Partial Decoupling Mechanism [Schedule 190, P.U.C. Or. 24, Sheet Nos. 190-1 through 190-3]. In order to address the more limited issue of usage declines resulting from implementation of conservation and efficiency programs, some state commissions allow recovery of lost delivery revenues associated with commission-approved programs outside of base rate cases. See, for example, Louisville Gas and Electric Co.'s Demand-Side Management Cost Recovery Mechanism [P.S.C. of Ky. Gas No. 6, Original Sheet Nos. 71-71.4] and Baltimore Gas and Electric Co.'s Conservation Surcharge Rider [P.S.C. Md. - G-9 (Suppl. 301)].
- AGA, "Patterns in Residential Natural Gas Consumption Since 1980," Feb. 11, 2000, page 7; "Patterns in Residential Natural Gas Consumption, 1997-2001," June 16, 2003, p. 5; and "Forecasted Patterns in Residential Natural Gas Consumption, 2001-2020," September 21, 2004, p. 3.
- The per customer increase in the Northeast between 1980 and 1990 is attributed to increased space heating penetration, due mainly to conversions from fuel oil-based heating during this time period.
- AGA, "Trends in the Commercial Natural Gas Market," October 23, 2002, p. 6.
- In their discussions of the classification of costs as customer, demand, or volume-related for cost study purposes, neither the National Association of Regulatory Commissioners' Gas Rate Design Manual, June 1989 (pp. 23, 35-48) nor the AGA's Gas Rate Fundamentals, 1987 (p. 142) treat any costs other than gas supply-related costs as volume dependent.
- For example, in the Missouri Gas Energy cost of service study in its last rate case (), I treated Account 871, Load Dispatching, an operations and maintenance account, as volume dependent. This expense amounted to less than 0.01 percent of the revenue requirement.
- Joint Statement of the American Gas Association and the Natural Resources Defense Council Submitted to the National Association of Regulatory Utility Commissioners, July 2004, p. 2.
- Too often cursory examination of summer month average usage provides the sole basis for rejecting a rate design that substantially reduces an LDC's reliance on volumetric revenue recovery based on "rate shock" concerns. Without close examination of customer-specific data throughout the year, such conclusions may be unwarranted. For example, certain low-use customers in the summer months may be among the high-use customers in the winter months.
- Consider two examples of rate design changes that move gradually away from volumetric-dependent revenue recovery. First, a two-step usage rate with a break at a low usage level could be introduced. By increasing the customer charge moderately and assigning a sizable portion of the volumetric revenue to the low-use, first block, the rate design will produce reduced reliance on volumetric rate recovery while minimizing rate shock concerns. Or, if adequate data are available, the residential class can be bifurcated into low annual use customers and remaining customers. The movement toward greater revenue recovery through fixed charges can initially be moderate for small annual use customers and significant for other customers without causing rate shock concerns under this bifurcated rate design. Over time, each of these rate designs can be modified toward greater reliance on fixed rate element revenue recovery.
Case Study 1 Missouri Gas Energy Experiences $3 Million Shortfall
A study of the Midwest confirms the declining residential and commercial usage trends.1 The study was prepared in conjunction with a Missouri Gas Energy (MGE) general rate case. MGE has more than 430,000 residential customers and more than 60,000 small commercial customers. Based on statistical analysis of MGE's residential gas usage from March 1994 through June 2003, it was determined that residential use per customer, after taking into account weather variations, fell between 1.7 Mcf per year and 2.5 Mcf per year across MGE's three geographic regions in Missouri.2 While based on a somewhat different time period, these results fall in the same range as the AGA's findings that show a 2 Mcf per year annual decline in weather-normalized residential per-customer use between 1997 and 2001 for customers in the Midwest Census region.3 In addition to MGE's declining residential usage pattern, it was found that small commercial per-customer usage, after taking into account weather variations, fell each year by 3.8 Mcf to 7.7 Mcf across MGE's three geographic regions in the 1994-2003 period.4
Declining residential usage trends may be due to both reduced weather-sensitive usage and non-weather sensitive, or base-load usage. Among the factors leading to reduced weather sensitive usage are improved appliance efficiencies, increased incidence of automatic thermostats, improved thermal efficiencies of homes, and increasing awareness of the value of conservation achieved through thermostat setbacks and insulation measures. Among the factors leading to declining base loads are improved water heater and clothes dryer efficiencies, greater incidence of pilotless water heaters and electronic-ignition stove tops, increased installations of water heater wraps, and lifestyle changes, including greater reliance on microwave ovens and convenience foods. Declining market penetration of various gas-burning appliances contributes to both falling per-customer base loads and weather-sensitive loads.
The MGE analyses confirm that both reduced weather sensitivity and declining base loads have contributed to the residential and small general service usage trends. Figure 1 demonstrates this result for MGE's Kansas City region's residential and small commercial customers. The figure shows declining base loads, approximated by average monthly usage in July and August, and declining weather sensitivity, measured by average November-March usage per customer per heating degree day.5
These usage declines are significant for Missouri Gas Energy. With the base rates in effect at the time the analyses were performed, Missouri Gas Energy would experience a delivery revenue shortfall of $3 million between the end of the test year in the rate case and the end of the first year after new rates became effective.
- In addition to my studies, other company-specific analyses demonstrate declining usage trends. Two residential studies, one involving Columbia Gas of Ohio and one involving Aquila, Inc. (Kansas), are examples. In his analysis of Columbia Gas of Ohio customer usage, William Gresham determined that residential base load usage per customer fell by 5.3 percent between 1980 and 2000 and heating load per customer fell by 6.6 percent during this same period (William Gresham, "Natural Gas Consumption Trends and Price Elasticity," Presentation to the Southern Gas Association Gas Forecasters Forum, October 2002). Gresham's residential statistical analysis also shows approximately a one percent per year decline in weather-normalized residential usage over the period 1990 through 2002. In his recent testimony in Aquila's Kansas rate case, Paul H. Raab noted that "From 1993 to 2003, residential usage in Aquila's Kansas service territory has dropped from 101 Mcf/year in 1993 to 79.3 Mcf/year in 2003, a reduction of 21%" (Direct Testimony of Paul H. Raab, State Corporation Commission of Kansas, , November 2004, p. 5, lines 19-21).
- Direct Testimony of F. Jay Cummings, Missouri Public Service Commission, , November 2003, p. 10. MGE's service area is split into the three geographic regions of Kansas City, Joplin, and St. Joseph. Separate residential regression analyses for each of MGE's three geographic regions produced adjusted R2 of at least 0.96, with weather and usage trend variables each significant at a 99 percent confidence level.
- AGA, "Patterns in Residential Natural Gas Consumption, 1997-2001" June 16, 2003, p. 5.
- Id. Separate small commercial (termed "small general service" in MGE's tariff) regression analyses for each of MGE's three geographic regions produced adjusted R2 of at least 0.94, with weather and usage trend variables each significant at a 99 percent confidence level.
- The HDD measure is typically used to measure the coldness of weather. The number of HDDs on a given day is the difference between 65o and the average of the high and low temperatures for the day (with no HDDs if the average temperature is 65o or higher). The number of annual HDDs is simply the sum of daily HDDs over the year. Data used in developing the figure comes from the usage trend analyses. The residential results are shown in tabular form on page 18 of the Surrebuttal Testimony of F. Jay Cummings, Missouri Public Service Commission, Case No. GR-2004-0209.
Case Study 2 CenterPoint Energy Arkansas Sees 1.2 Mcf/Year Fall in Residential Use
The second study was prepared in conjunction with a CenterPoint Energy general rate case in Arkansas. This study covered the period from January 1987 through June 2004. The residential statistical analysis showed that residential use per customer, after taking into account weather variations, fell approximately 1.2 Mcf per year over this period.1 These results fall in the same range as those found in AGA's studies. For residential customers in the South Census region that includes Arkansas, AGA found a 1.3 Mcf per year decline in per customer usage between 1990 and 1997 and a 1.2 Mcf annual decline between 1997 and 2001.2
For CenterPoint Energy small commercial customers, my analyses revealed a 7.3 Mcf per year usage decline, after taking into account weather variations, over the 1994-2003 period.3 With more than 380,000 residential customers and more than 48,000 small commercial customers in Arkansas, these trends have a significant impact on CenterPoint Energy's delivery service revenue collections.
Consistent with findings from other studies, CenterPoint Energy's Arkansas residential usage trend involves both base load and weather-sensitive load reductions. Figure 1 shows a consistently declining residential base load, approximated by average monthly usage during the months of July through September, over the 1987-2003 period.4
The reduced weather-sensitivity of CenterPoint Energy's residential customers is revealed by comparing estimated weather-sensitive usage in recent years to years in the last decade with comparable weather, , pairs of mild, approximately normal, or cold years. Figure 2 shows that CenterPoint Energy's Arkansas residential customer weather-sensitive usage consistently was lower in each year in this decade compared to a year with comparable weather in the 1990s (with the data labels showing annual estimated weather-sensitive usage (in Ccf) per customer).5
The Arkansas residential usage trend, combining both base load and weather-sensitive usage, is summarized in Figure 3. This figure provides estimated weather-normalized annual usage per customer in each year from 1987 through 2004.6
The usage of CenterPoint Energy's small commercial customers in Arkansas show a similar generally declining usage pattern, both base load and weather-related, as summarized in Figure 4.7
- Direct Testimony of F. Jay Cummings, Arkansas Public Service Commission, Docket No. 04-121-U, December 2004, p. 34. The best-fit residential regression produced an adjusted R2 of 0.98, with weather and usage trend variables each significant at a 99 percent confidence level.
- AGA, "Patterns in Residential Natural Gas Consumption, 1997- 2001" Feb. 11, 2000, p. 7 and AGA, "Patterns in Residential Natural Gas Consumption, 1997-2001" June 16, 2003, p. 5.
- The best-fit regression for the small commercial class produced an adjusted R2 of 0.95, with weather and usage trend variables each significant at a 99 percent confidence level (Direct Testimony of F. Jay Cummings, December 2004, page 36). CenterPoint Energy's current small commercial class definition became effective in January 1993. As a result, 1994 is the earliest year ended in June that is included in the analysis for this class.
- , Exhibit FJC-7.
- Data used in the figure are contained in Id., p. 35.
- Estimates developed using the methods explained in Id., p. 35.
Case Study 3 Texas Gas Service Finds Usage Declines Despite Small Base
Declining usage trends appear even in a service area with a small customer base. Take for example Texas Gas Service Co.'s South Jefferson County Service Area, an area near the Texas gulf coast with approximately 30,000 residential customers.1 Based on the period from July 1995 through December 2002, the statistical analyses showed that annual residential use per customer, after taking into account weather variations, fell approximately 1.05 Mcf per year over this period.2 The Texas Gas Service result is quite similar to, although somewhat smaller than, the finding for the other LDC in the South that was examined and results reported in AGA in Case Study 2 (). Based on the data used to establish this trend and the measure of normal weather used in the rate case, Figure 1 provides estimates of South Jefferson County, Texas residential weather-normalized annual usage per customer in each year from 1996 through 2002.
- In addition to analyzing usage patterns in this Texas area, I conducted an earlier study of annual residential per customer usage in Southern Union Gas Company's (now Texas Gas Service Co.'s) El Paso Service Area, with approximately 154,000 residential customers at the time. I found a statistically-significant residential usage per customer trend that showed a 1.33 Mcf per year decline over the twenty-year period of 1977 through 1996 ().
- Texas Gas Service Company, a division of ONEOK, Inc., has a number of distinct services areas throughout Texas. The South Jefferson County Service Area is composed of the Cities of Port Arthur, Groves, Nederland, and Port Neches and surrounding unincorporated areas. Direct Testimony of F. Jay Cummings, Railroad Commission of Texas, GUD No. 9465, November 2003, p. 8, lines 7-9. The best-fit regression for the residential class produced an adjusted R2 of 0.95, with HDD and usage trend variables each significant at a 99 percent confidence level. The regression analyses for the commercial class produced negative, but not statistically meaningful, usage trends. This result is not surprising because there are only 1300 commercial customers in the service area, and all commercial customers, regardless of size, are included in this single class.
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.