Capacity markets

The short version

An energy-only market pays generators only when they produce power. That creates a problem: during the 8,700 hours a year when prices are low, a peaker plant earns almost nothing — yet it needs to be available for the 50 hours a year when demand spikes and the grid needs it most. Without additional compensation, rational investors won't build or maintain those plants.

Capacity markets solve this by paying resources separately for the option to generate. A generator that clears a capacity auction receives a stream of $/MW-day payments — an availability fee — in exchange for a commitment to be physically available during declared emergencies. It's the grid's insurance premium, paid in advance.

How the auction works

Each ISO runs its capacity auction on its own schedule, typically one to three years before the delivery period. The process starts with a demand curve — a relationship between the amount of capacity procured and the price the ISO is willing to pay. More capacity means lower prices; approaching the minimum requirement, prices rise steeply.

Generators submit offers specifying how much capacity they can provide and at what price. Demand response providers, behind-the-meter storage, and increasingly virtual power plant aggregations can also submit offers. The ISO solves a procurement optimization, accepts resources from lowest to highest cost until the demand curve is satisfied, and sets a single clearing price paid to all accepted resources.

In PJM, the Base Residual Auction (BRA) clears three years ahead. Capacity Performance rules, introduced after the 2014 Polar Vortex, impose steep penalties on resources that fail to perform and redistribute those payments to resources that do. In ISO-NE, the Forward Capacity Auction (FCA) clears three years ahead with a similar performance requirement structure. NYISO runs monthly ICAP auctions alongside a spot market.

ICAP vs. UCAP

Not all megawatts are equal in a capacity market. ICAP (installed capacity) is a resource's nameplate rating. UCAP (unforced capacity) discounts that rating by the historical probability of forced outages — a gas plant with a 6% forced outage rate contributes only 94% of its nameplate as UCAP. This matters because the capacity market is ultimately about reliability, and a resource that is frequently unavailable provides less reliability value than its nameplate suggests.

Markets without capacity auctions

CAISO and ERCOT use different approaches. CAISO relies on resource adequacy (RA) requirements — load-serving entities must procure sufficient capacity to cover their peak load plus a reserve margin, demonstrated through bilateral contracts rather than a centralized auction. ERCOT is an energy-only market; the Operating Reserve Demand Curve (ORDC) adds an adder to real-time prices when reserves are tight, theoretically incentivizing investment through high energy prices during scarcity events.

What this means for energy buyers

Capacity costs flow through to large C&I customers as a component of their retail electricity bill. The allocation is typically based on coincident peak demand — your load during the grid's peak hour(s). In PJM, this is the single highest-load hour of the year; in ISO-NE and NYISO, it's typically based on your contribution to the top peak hours.

This creates a powerful incentive: reducing your load during those specific peak hours — through demand response, on-site generation, or pre-cooling — directly reduces your capacity cost obligation for the following year. For large industrial facilities, peak shaving strategies can reduce capacity tag obligations by 20–40%, generating meaningful bill savings.