Demand Charges: What Do They Really Measure? By John Kelly Demand charges have been staples in the electric utility industry for decades. They are attempts to allocate portions of capital expenditures, most significantly those for generating facilities, to individual large customers or groups of customers. The charges are supposed to represent payments for relative amounts of generating capacity for which customers are responsible. But what do demand charges really indicate? What do they measure? How reliable are they as a guide for understanding costs and designing rates? The conventional justification for demand charges may seem reasonable when viewed casually, but a closer look reveals serious problems. First, such charges are the product of a futile exercise—an attempt to allocate common costs in a meaningful, economic way (See “The Problem with Cost Allocation: The Heart of the Matter,” Public Power, March-April 2004). The general logic of demand charges does not hold up under scrutiny, nor do the assumptions that underlie their calculation. Demand charges are typically assigned to individual customers or to groups of customers (customer classes) based on the proportion of kilowatt demand those customers or customer classes contribute to a utility’s peak. The capital costs of generating facilities are allocated based on such proportions and then assessed in terms of dollars per kilowatt. There are several variations of this calculation (e.g., using different peak periods or combination of them) but the basic result is the same: consumers are assigned particular, fixed kilowatt demand charge rate 40 MAY-JUNE 2004 (e.g., $5/kW, $8.50/kW). Customers a utility’s peak demand: 20 percent; 30 usually pay this assigned charge for at percent; and 50 percent. The demandleast a year. More typically, though, the charge rate structure would use these periods between adjustments are much proportions as major factors in assignlonger, often several years, before an- ing fixed costs, justifying the allocations other cost study is done to estimate by contending that these customers demand charges. The demand charge caused these relative amounts of genrate is multiplied by the amount of a erating capacity to be built. But what customer’s maximum monthly kilowatt happens if a cost analysis is performed demand(which usually varies) to arrive the following year and the relative proat the total demand-charge payments portions change to 25, 35, and 40 (e.g., $5/kW multiplied by 50 kW percent? Did the causal responsibility equals $250 in total-demand charge change? And what happens in year payments). The resulting pattern of three if the customer that contributed rates can be characterized as a “de- 40 percent to the system peak leaves and is replaced by a new customer mand-charge rate structure.” Calculating a significant portion of the cost of electric Calculating a significant portion service using de- of the cost of electric service mand charges is using demand charges is like like using one frame of an 8,760- using one frame of an 8,760frame film to tell a frame film to tell a story. story. Included in the frame, frozen in a brief period of time, typically one whose proportion of peak load is also hour, are a utility’s balance sheet and 40 percent? Is it reasonable to say that selected operating costs, juxtaposed. A the new customer caused, or was in any utility may have millions of dollars in- way responsible for, the construction of vested in generating facilities and existing generating facilities, which during its system peak there are hun- were likely built many years before? All dreds of possible amounts individual that can be said is the descriptive fact customers or customer classes could that at a particular time a customer or have contributed to that peak. These customer class happened to use a ceramounts will then be used to determine tain proportion of the existing a rate (dollars per kilowatt) at which megawatt capacity. It is unrealistic to assume that what customers will be billed for subsequent happens in one hour during a particular consumption. The flawed assumptions underlying year, or some other period, is represenof demand charge calculations are il- tative of what happened in past years, lustrated by a simple example. Assume what’s happening currently, and what there are only three utility customers will occur in the future. Because parwho accounted for these proportions of ticular customers or customer classes PUBLIC POWER www.APPAnet.org MAY-JUNE 2004 41 Demand Charges happen to have certain kilowatt demands in one film frame does not mean they were responsible for related proportions of capital investments in generation facilities. It is unlikely that the reasons for, and decisions to build existing generation facilities are even on the same film reel. Customers come and go over the productive life of generating facilities. To say that a particular customer or customer class was responsible for a This example becomes more troublesome if the utility decides to offer a discounted rate to attract the customer. On one hand, some will argue that it makes sense to give the potential customer a special discount because the new load will help pay for existing capacity. Yet others will say the conventional view that the potential customer has in some way caused, and is consequently responsible for, construction of existing peaking capacity. If the logic and practicality of deDemand charges often mand charges make sense, then why penalize customers with low aren’t they used in load factors and reward those o t h e r c a p i t a l with high load factors. intensive industries such as restaurants, hotels, phone comparticular increment of generating ca- panies, and airlines? Hotels in major pacity is virtually impossible, unless cities that cater primarily to business facilities were dedicated under some travelers may let rooms at $200 or contractual arrangement. more a night during peak times, then The difficulties with and contradic- drop the price to $100 during nontions encountered in the use of demand peak times. A person who books a charges are obvious when a large in- room during peak season is not asdustrial customer is urged to locate in a signed a demand charge (based upon community. If the company will con- some estimate of the portion of addisume a relatively large share of the tional facilities needed to meet peak utility’s peak demand, can it be said to demands) that he or she carries during have caused that peak? Clearly not. the rest of the year, and that will be collected if the person returns during non-peak times. Summer resort restaurants that remain open during the winter do not assign demand charges to customers for the extra space, chairs, tables, dishes, silver and facilities required to meet peak summer demand. If customers return in the winter for a meal, they are not expected to pay the regular off-season price of a meal plus a demand charge that was allocated to them during the summer. In the airline industry, summer and holiday travelers are not assigned demand charges for the extra planes required to meet peak demands. They are not allocated peak costs and then expected to pay them when they travel in off-peak periods. Off-peak travelers pay the cost of operating the plane they are 42 MAY-JUNE 2004 traveling on plus any mark-up the airline can impose to make the flight as profitable as possible. This usually means low fares that are close to operating costs (including maintenance costs and economic depreciation of the plane). F rom a historical perspective, even one of the early, leading proponents of demand-charge rate structures conceded “that it could be easily shown” that peak responsibility was “not the theoretically correct measure for the demand charge” (“Engineers and Economists: Historical Perspectives on the Pricing of Electricity,” Technology and Culture, William J. Hausman and John L. Neufeld, January 1989). Hausman and Neufeld are quoting engineer Hugh Eisenmenger. They go on to note that in 1921 one of the early critics of demand charges, economist George Watkins, argued that customers with large individual peaks were not necessarily more expensive to serve than other customers. Assuming that they were was a serious theoretical and practical flaw in the demand charge. Demand charges often penalize customers with low load factors and reward those with high load factors. The conventional thinking in Watkins’ time, as it is today, is that such an arrangement is appropriate. It promotes efficiency because a high load factor suggests a relatively smooth pattern of consumption and thus reduces the average demand costs per kilowatt. Hausman and Neufeld noted that Watkins took issue with this view and with its suggestion that “what is good for the utility should be rewarded in individual customers.” Watkins contended: It is easy, at this point, to let one’s reasoning go astray by identifying the interests of the company in building up a good load for itself with a policy of favoring consumers with good individual load factors. The acquisition of a new consumer with an individual load factor better than that of the company must, it is true, tend to raise PUBLIC POWER Demand Charges So why are demand charges so 20th century at about the same time prevalent? The answer is twofold, “utilities faced a serious competitive partly historical accident and partly challenge in supplying electricity to economic motivation (“The Curious industrial users in the self-generation of Economics of the Electricity Demand electricity” by these users, Neufeld Charge,” Public Power, John Neufeld, said. The demand-charge rate strucMay-June 1999). Rate structures rest- ture gave the electric utility industry a ing on sounder conceptual grounds were proposed in the early days of the elec- The demand-charge rate tric utility industry. structure served double duty Neufeld found that “much of the early debate on rate during the years 1907-1914, structures was fueled by when most states began the desire to sell meters.” regulating utility rates. One of the particularly effective salesmen was an Englishman, Arthur Wright, who had means to set prices based on the value the backing of Samuel Insull, one of the of service to industrial customers rather most prominent individuals in the U.S. than what it cost utilities to serve such electric power industry. Wright held customers. If a customer’s maximum patents on the first practical demand- demand was 10 kilowatts and the comcharge meter and advocated a variation pany could afford to purchase and operate its own generating equipment, of the demand charge rate structure. All this happened during the early the utility would offer a demand charge intended to approximate the cost of the customer operating a 25-kilowatt generator of its own. If another customer had a maximum demand of 50 kilowatts, the utility would structure a demand charge based on that amount. The intention was not to set rates based on a utility’s actual cost of serving various customers, but to pricediscriminate to retain and attract customers and to maximize profits. The demand-charge rate structure served double duty during the years 1907-1914, when most states began regulating utility rates. Manufacturing companies worked to prevent utilities from practicing the price discrimination that occurred in private negotiations of rates. State regulation required that rates could no longer be determined by negotiation and had to be published in publicly available schedules. But utilities were successful in holding on to demand-charge rate structures and these structures provided “an exquisite device for automatically supplying discounts, where needed,” or, more simply, for practicing price discrimination. ● 44 MAY-JUNE 2004 PUBLIC POWER