Paying Large Customers Not to Use Electricity
It is frequently cheaper to pay somebody to consume less at peak than to build generation that runs for a few hours a year. Demand response markets pay for that reduction, and measuring it requires a counterfactual.
The Cheapest Megawatt Is the One Nobody Uses
Electricity demand peaks for a small number of hours each year, typically on the hottest summer afternoons. Meeting those peaks requires generation capacity that sits idle nearly all the time.
Building a plant to run for fifty hours a year is extraordinarily expensive per unit of energy delivered. The alternative is to reduce demand during those hours instead.
Demand response is the market mechanism for that: participants commit to reduce consumption when called, and are paid for the commitment and for the reduction.
What Gets Paid For
Programmes divide into two families with different economics.
| Type | Payment Basis | Called When |
|---|---|---|
| Capacity or reliability | Paid to be available, plus for events | System is short of capacity |
| Economic or price responsive | Paid for energy reduced | Wholesale prices are high |
Reliability programmes are the larger category by value. A participant commits to a defined reduction and receives a capacity payment for the commitment regardless of whether an event occurs, plus energy payments when it does.
That structure means the revenue is largely an availability payment, which is what makes participation attractive to a business that may only be called two or three times a year.
The Measurement Problem
Paying for generation is straightforward, since a meter records what was produced. Paying for reduction requires knowing what would have been consumed without the event, which is unobservable.
The answer is a baseline, an estimate of counterfactual consumption, typically calculated from the customer usage on similar recent days, sometimes with a same day adjustment using consumption in the hours before the event.
Every baseline method is gameable, and the direction of the gaming is obvious. A customer that raises consumption on the days used to construct the baseline, or immediately before an event, appears to reduce more than it did.
Programme rules address this with adjustment caps, exclusion of prior event days, and randomised or symmetrical baseline windows. None of it is perfect, and baseline design is where the integrity of the whole product sits.
Demand response pays for something that did not happen, measured against an estimate of what would have happened. That is an unavoidable feature of the product and it is why the rules are complicated and why the payments are audited.
Who Participates
Historically the participants were large industrial and commercial customers with genuinely interruptible processes: cold storage that can coast for two hours, industrial furnaces with thermal inertia, water pumping that can be rescheduled, and buildings whose cooling can be shifted.
The common feature is a process where timing is flexible and output is not lost, only moved. A steel mill that must halt a heat loses the batch, and no capacity payment compensates for that.
Residential participation arrived later through aggregation, principally smart thermostats and water heaters, and it now constitutes a meaningful share of enrolled capacity in several markets.
The Legal Fight Over Compensation
A genuinely important dispute concerned how much a demand response resource should be paid in wholesale markets.
A federal order established that demand response should be compensated at the full wholesale energy price when it is cost effective, treating a megawatt reduced as equivalent to a megawatt generated.
Generators objected, arguing that a customer reducing consumption already avoids paying for the electricity, so paying the full wholesale price in addition compensates them twice. The rule was challenged and ultimately upheld by the Supreme Court in 2016, which confirmed both the regulator authority over the practice and the compensation approach.
The economic argument has not gone away. Whether the double payment criticism is correct depends on whether you regard the avoided purchase as a benefit the customer already receives or as the cost of the reduction they undertook, and reasonable economists differ.
What Limits It
Three constraints keep the resource smaller than the theoretical potential.
Event fatigue. Participants tolerate a small number of interruptions and enrolment falls if events become frequent. Programmes cap the number and duration of events, which caps the value.
Snapback. Consumption deferred is frequently consumed immediately afterward, as buildings recool and processes catch up, which can create a secondary peak and reduces the net benefit.
Verification cost. Metering, settlement, and baseline calculation carry administrative cost that makes small participants uneconomic without aggregation.
How It Fits the Changing Grid
The traditional case was avoiding peaking plants. The emerging case is different: as generation shifts toward resources whose output varies with weather, the system needs flexibility across more hours and in both directions.
That has produced interest in paying customers to increase consumption when generation is abundant and prices are negative, which is the mirror image of the original product and requires the same baseline machinery.
The Bottom Line
Demand response pays for reduced consumption because avoiding a peak is far cheaper than building capacity to serve it, and the resource is genuinely valuable to a grid with limited flexibility. Everything difficult about it follows from paying for a counterfactual, which makes baseline design the load bearing element and the place where the product can be gamed. The compensation question was litigated to the Supreme Court and settled legally without settling the economics.