Demand charges have existed since nearly the beginning of the electric industry. Although utilities often favor them, economists have continually questioned whether these rates — applied to an individual customer’s maximum short-term usage in a billing period — are an efficient form of pricing. This paper does a deep dive on demand charges and cost causation. Like many analysts from the past, the authors find that demand charges have made sense only as a proxy and are not a general solution for shared capacity costs. Furthermore, the changes occurring with a modern grid are undermining the conditions that made such a proxy reasonable.

With a few narrow potential exceptions, the technological capabilities of the 21st-century electric system are rendering demand charges obsolete. The authors explore three main reasons for this.

First, unlike time-varying energy pricing, demand charges fail to properly recognize the essential role that timing plays in driving shared system costs. Demand charges penalize customers for usage at times that do not impose particularly high costs and encourage them to waste effort and money shifting loads off their own maximum hour — and sometimes onto hours when system load and thus costs are higher. A flat individual customer load shape may not, in fact, be best for the system. With the high penetrations of variable renewable generation in today’s electric system, load that can respond to swift changes in the availability of supply can be cheaper to serve than unvarying loads.

Traditional demand charges also fail to reflect the degree to which capacity costs in the modern electric system are incurred to meet energy needs rather than demand. Where once a megawatt of capacity built for peak demand could be used equivalently year-round, the mix of resource types in today’s system changes the calculation. A significant portion of capacity investment today is designed to reduce energy costs and should be charged on that basis.

Finally, even a demand charge that recognizes to a certain extent the time-dependent nature of costs — a peak window demand charge — retains many of the arbitrary and inefficient characteristics of traditional demand charges. Dramatic decreases in the cost of sophisticated metering, which enable a peak window demand charge, can also facilitate superior time-varying rate options.

After assessing the shortcomings of demand charges, the authors ask whether any circumstances exist in which this form of pricing might be appropriate. The cases they identify are: (1) site infrastructure for individual customers, (2) risks related to customer variability at peak times and (3) timer peaks. Even here, though, demand-based pricing would only be a second-best approximation of a more efficient time- and location-based rate structure.