Cost of Service Ignores Load Factor

Fortnightly Magazine - February 1 1995

In his recent article, "Cost-of-Service Studies: Do They Really Tell Us Who's Subsidizing Who?" (Nov. 15, 1994), Mark Quinlan proposes an alternative cost-of-service methodology. He claims that under current cost-allocation methods (and given adequate capacity to meet demand) a rate class with increasing sales subsidizes a rate class with decreasing sales. This may not be true, depending on how costs have previously been assigned to rating periods, classified, and allocated.

As the article notes, between 1984 and 1991, United Illuminating Co.'s (UI's) industrial sales have decreased by 25.4 percent while its residential sales have increased by 12.7 percent. Contrary to Mr. Quinlan's assertion, however, a higher proportion of costs has not been supported by residential sales. UI has never used major customer classifications (residential, commercial, industrial) to establish its cost-of-service rate classes; instead, UI uses rate schedules based on customers' load characteristics. Therefore, what Mr. Quinlan refers to as large industrial rate classes are actually a mix of commercial, industrial, and a few high-use residential customers. These rate classes, general service time-of-use (GST) and large power time-of-use (LPT), have increased usage much more over time than UI's primary residential rate (R) class. From 1983 to 1993, GST's usage increased by 1,302,705 percent, from 44 megawatt-hours (Mwh) to 573,234 Mwh. LPT usage increased 36.3 percent, from 1,305,316 to 1,779,278 Mwh. In this same time period, R usage increased by 23.7 percent, from 1,143,905 to 1,414,781 Mwh.

During this time period, three of UI's cost-of-service rate classes declined in usage; under Mr. Quinlan's theory, then, they were being subsidized by other rate classes. Let's look at these three rate classes and examine why their usage decreased.

EES North America

The street lighting (SL) rate class reduced usage by 8 percent, from 59,202 to 54,841 Mwh. Although there are many more lights out there, this decline can be traced to extensive conservation efforts, which have converted virtually all lights from mercury vapor to high-pressure sodium lighting. I doubt whether regulatory commissions across the county would be in favor of penalizing this class for past conservation efforts.

Usage in the general service rate class has decreased by 18.5 percent, from 1,105,277 to 900,479 Mwh, because higher-load customers in this group migrated to GST. Should we penalize the customers who have remained in this class? I don't think so.

Rate class A's (residential heating and offpeak) usage has declined by 17.7 percent, from 497,149 to 409,256 Mwh, because offpeak pricing is too high. Why? Because under current cost-of-service capital substitution and averaging of costs assigned to rating periods methods, this class's rate of return (ROR) actually comes out lower than standard rate class R. This class has much higher usage per customer than rate class R-10,567 Kwh annually, compared to 6,050 Kwh-and is primarily an offpeak load. We would argue that it is the inappropriate cost-of-service method that creates the lower ROR in this class. Yet, if we adopt Mr. Quinlan's method for classes with declining sales, we will continue to increase the prices in this class each time more customers are chased away by the last price increase, until none are left.

From 1983 to 1993, UI's average system price increased 8.3 percent-from 10.66 cents per Kwh (›/wh) to 11.55›/Kwh. SL rates increased 66.7 percent-from approximately 15 to 25›/Kwh. Yet the last cost-of-service study (COSS) submitted by UI and approved by the Connecticut DPUC showed that this class's ROR that was not only below system average, but actually negative (1.269 percent) when compared to the system average of 11.05 percent. Now, any reasonable person might ask how a class of service that is completely off peak, operates with 100-percent load factor, and pays rates 116.5 percent higher than average could have a negative ROR. The answer lies not only in the use of the misguided capital substitution average and excess cost-allocation methodology, but more in the way the DPUC has directed that costs be asigned to rating periods in UI's time-differential COSS.

There is virtually no possibility of UI peaking at any time other than noon to 5 pm, Monday through Friday, during the months of June through September. UI has designated the hours of 10 am to 6 pm, Monday through Friday, during the months of June through September as the onpeak period. This amounts to approximately 704 hours, or 8 percent of the total hours in the year. Shoulder is comprised of 2,080 hours, or 23.8 percent, and offpeak accounts for 5,976 hours, or 68.2 percent. UI proposed using the base, intermediate, and peak (BIP) method of assigning costs to these rating periods. For example, a base load plant that ran continuously would be assigned one-third to base hours, one-third to intermediate hours, and one-third to peak hours. The DPUC method assigns a percentage of annual hours in each rating period. Therefore, 8 percent of the costs of base load plants are assigned to the onpeak period, and 68.2 percent were assigned to the offpeak period. This has a devastating effect on offpeak loads.

Cost-of-service methods should realistically link the actual full costs of service to the unique load characteristics of the customer class and properly recognize the higher costs of serving short-duration peak loads. In today's changing environment, even a COSS that uses correct allocation methods is probably no longer relevant in designing rates. Value of service, which recognizes class inelasticities, is much more important.

Jim Lundrigan

Pricing Manager

The United Illuminating Co.

New Haven, CT


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.