Capacity Accreditation and ELCC: Why the Credit for Solar and Storage Keeps Getting Cut
Between the 2027/2028 and 2028/2029 Base Residual Auctions, PJM cut the ELCC class rating for onshore wind from 41% to 28%. Nothing about any wind farm changed. The turbines were the same, the resource was the same, the interconnection was the same.
On a 200 MW project, that is about $3.3m a year of capacity revenue. And when you decompose it, roughly 95% of the loss came from the accreditation change and only 5% from the clearing price — the price everyone watches, reports and models.
The previous post ended on the observation that accreditation moves the PJM capacity line more than the clearing price ever does. This post is about why it moves, why the direction is consistently downward for the technologies being built, and how to model a revenue line whose defining variable is recalculated annually.
ℹ️ Note: Class ratings are published per delivery year and change. Figures here are as published for the 2027/2028 and 2028/2029 auctions and should be re-checked against the current publication rather than carried forward.
What Is ELCC?
Effective Load Carrying Capability measures how much additional load a resource lets the system serve at the same reliability level. It answers a reliability question, not an energy question: not how much a plant produces, but how much of the system's risk it actually removes.
That distinction is the whole subject. A solar farm can produce an enormous number of megawatt-hours and remove very little loss-of-load risk, if the hours it produces in are not the hours the system is short.
The 2028/2029 Class Ratings
| Class | 2028/2029 ELCC |
|---|---|
| Fixed-tilt solar | 5% |
| Tracking solar | 7% |
| Onshore wind | 28% |
| Offshore wind | 47% |
| 4-hour storage | 55% |
| 6-hour storage | 65% |
| 8-hour storage | 67% |
| 10-hour storage | 75% |
| Landfill intermittent | 55% |
| Hydro intermittent | 37% |
| Demand resource | 65% |
Read the solar rows first, because they are the ones people find hardest to believe. A 200 MW tracking solar plant is accredited at 14 MW. At $325/MW-day that is roughly $1.66m a year of capacity revenue against a project that might cost $250m to build.
Then read the storage rows, which are the more interesting structure. Storage accreditation is not a single number — it rises with duration, from 55% at four hours to 75% at ten. Duration, not power, is what the accreditation framework is buying.
What Changed in One Year
| Class | 2027/2028 | 2028/2029 | Change |
|---|---|---|---|
| Fixed-tilt solar | 7% | 5% | −29% |
| Tracking solar | 8% | 7% | −13% |
| Onshore wind | 41% | 28% | −32% |
| 4-hour storage | 58% | 55% | −5% |
A third of onshore wind's capacity accreditation, removed in one publication cycle. That is a larger move than any capacity clearing price change in the same period, applied to the same assets, and it arrives from a planning process rather than from an auction.
Why Is Solar Accredited So Low?
Because the risk hours moved, and solar cannot follow them.
Capacity value is earned in the hours when the system is closest to shedding load. Historically those were summer afternoons, which is exactly where solar output sits — and solar accreditation was correspondingly high when penetration was low.
Adding solar shifts the risk. As MISO's analysis describes it, additional solar capacity shifted risk hours slightly later in the day, to the point where both wind and solar resources had a lower capacity value in summer months. The afternoon peak is covered; the evening ramp after sunset is not; and the marginal reliability hour migrates to a time when a solar plant produces nothing.
This is the mechanism people find counterintuitive and it is worth stating directly: solar's accreditation fell because solar worked. Enough of it was built to solve the summer afternoon problem, and once that problem is solved the next increment is not paid for solving it again.
The Migration to Winter
The second, larger force, and it is the one that makes the trend structural rather than cyclical.
Ascend Analytics' PJM work finds that system stress is concentrating increasingly in winter — by 2031, nearly 90% of loss-of-load expectation risk occurs in winter months, against about 13% in summer.
That reallocation is devastating for a summer-peaking resource. Winter reliability events in PJM are cold-morning and cold-evening events: long, dark, and often calm. A solar plant contributes nothing to a 7am January event. A four-hour battery contributes for four hours of an event that may last considerably longer — Ascend notes that prolonged winter stress exceeds storage discharge duration, making longer-duration assets more valuable.
That single sentence explains the whole duration ladder in the class ratings table. The 55/65/67/75 progression is not a technical curiosity. It is the accreditation framework pricing the observation that winter events are long.
Why Does Accreditation Fall as Penetration Rises?
Because ELCC is marginal, and the marginal unit is worth less than the ones before it.
As a resource class becomes a larger share of the supply stack, its incremental reliability contribution decreases. The first tranche of a technology covers the hours it is good at. The second tranche covers the same hours, which are now already covered. The reliability value of each additional unit falls, and because the class rating applies to the whole class, it falls for the existing fleet too.
The forecast magnitude is severe. In Ascend's 2026 base case, the ELCC value of four-hour battery storage falls from 57% to 20%, with eight-hour units seeing comparable reductions. Over the same horizon solar and wind capacity both grow roughly fivefold while their reliability contributions fall.
Two consequences deserve to be stated plainly.
Accreditation is a shared, not a project-specific, variable. A project's capacity revenue depends on how much of its own technology everybody else builds. No amount of diligence on a single project reveals it, and no contract allocates it.
It is not symmetric. There is no mechanism by which a class rating rises back toward its earlier level unless the build-out reverses. Modelling accreditation as mean-reverting is modelling it backwards.
Average or Marginal? The Choice Sets Both the Level and the Slope
A methodological point that decides more than it looks like it does.
There are two defensible ways to measure a resource class's capacity contribution. Average capacity credit asks what the existing fleet of a technology contributes, divided across that fleet. Marginal capacity credit asks what the next increment contributes.
While a technology is scarce the two are close. As it saturates they diverge sharply, and always in the same direction: the average stays high because it includes the early units that were genuinely valuable, while the marginal figure collapses because the next unit adds to hours already covered.
Both answer real questions. Average credit describes what has been built. Marginal credit describes what should be built next, and is the correct signal if the purpose of the payment is to steer investment.
The consequence for a financial model is that the choice of methodology sets not only the level of the rating but its rate of change. An average-based construct declines slowly, because each new unit is diluted across a growing fleet. A marginal-based construct declines fast, because it tracks the frontier. The 41% to 28% move in a single cycle is what a marginal framework looks like when a technology is saturating quickly.
That is also why a project cannot argue its way out of the number. The rating is not a judgement about this wind farm; it is a statement about what the next wind farm would contribute. A project with a superb capacity factor in the right hours receives the same class rating as a mediocre one, unless the market offers a resource-specific accreditation path — and most do not.
Capacity Interconnection Rights Cap What You Can Sell
A separate constraint, often discovered late, and it is not the same thing as accreditation.
A project can only offer capacity up to its Capacity Interconnection Rights — the quantity established through the interconnection study process. Accreditation determines the percentage; CIRs determine the base the percentage applies to. Both have to be right for the revenue line to be right.
Ascend's analysis notes that capacity interconnection rights affect wind and solar ELCC values by at least five percent, which is material against ratings that are already in single digits for solar.
Two practical traps follow.
An energy-only interconnection produces no capacity revenue at all. Where a project elects an energy resource interconnection service rather than network resource interconnection service — often to get through the queue faster or to avoid network upgrade costs — it may have no CIRs to sell against. That is a rational trade, but it deletes the capacity line entirely, and a model carrying capacity revenue for a project without the corresponding rights is modelling revenue the project cannot earn.
CIRs are set on nameplate at a point in time. A repowering, an uprate or a change in configuration does not automatically bring more rights. The interconnection process governs, on its own timetable, which is the subject of the next two posts in this series.
The general instruction is to verify the CIR quantity from the interconnection agreement, not from the nameplate in the model.
MISO Is Doing the Same Thing Differently
Worth knowing, because the direction is identical even though the methodology is not.
MISO is moving wind and other resource classes to a Direct Loss of Load methodology from Planning Year 2028–2029, accrediting resources on their availability during critical hours in MISO's probabilistic loss-of-load model. The method heavily weights performance during the top 65 "at-risk" hours in each season.
Sixty-five hours out of roughly 2,190 in a season. Under DLOL, a resource's entire capacity accreditation turns on its availability in about three percent of the hours — and specifically the three percent that are hardest.
MISO has been publishing indicative DLOL results ahead of each Planning Resource Auction since PY 2025–2026, precisely so participants can see the change coming before it binds. The effect reported so far is that the new method benefits storage more than solar, which is the same conclusion PJM's class ratings reach by a different route.
Two markets, two methodologies, one answer: pay for availability in the hours of actual risk, and those hours are moving away from solar.
The convergence is itself informative. When two independent system operators, running different models against different fleets under different state policies, arrive at the same conclusion by different routes, it is unlikely to be a methodological artefact that better advocacy will reverse. A developer treating a class rating cut as an error to be litigated is generally reading it wrongly. The rating is measuring something real — that the hours the system is short are no longer the hours a solar plant produces — and the arithmetic will keep saying so until the fleet changes shape.
There is also a timing asymmetry worth planning around. PJM publishes preliminary class ratings for delivery years well beyond the auction currently being run, and MISO has committed to publishing indicative DLOL results before each Planning Resource Auction. The forward view is therefore available years ahead of the year it binds — which means a model carrying a flat rating is not missing information that does not exist. It is ignoring information that has been published.
How Do You Model Accreditation in Excel?
As a declining series with an explicit assumption, never as a constant — and decomposed, so the model says which variable moved.
The decomposition that makes the case
Assumptions, labelled as such:
Onshore wind project 200 MW
ELCC 2027/2028 41%
ELCC 2028/2029 28%
Clearing price 2027/2028 $333.44/MW-day = $121,706/MW-yr
Clearing price 2028/2029 $325.00/MW-day = $118,625/MW-yr
Capacity revenue 2027/28 200 × 41% × 121,706 = $9,979,892
Capacity revenue 2028/29 200 × 28% × 118,625 = $6,643,000
-----------
Change = ($3,336,892)
Now split it:
Accreditation effect (56 − 82) MW × $121,706 = ($3,164,356) 94.8%
Price effect 56 MW × ($118,625 − 121,706) = ($172,536) 5.2%
-----------
($3,336,892)
Ninety-five percent of the loss is accreditation. A model that tracks the clearing price carefully and holds the class rating flat has monitored the variable that explains five percent of the outcome.
The forward case that must be run
Base ELCC held at 28% → capacity revenue $6,643,000
Decline ELCC 28% → 20% → capacity revenue $4,745,000
Decline ELCC 28% → 15% → capacity revenue $3,559,000
Floor ELCC 28% → 7% → capacity revenue $1,661,000
The 20% case is not a stress test. It is the level Ascend's base case projects for four-hour storage, and it is barely below the 28% wind figure that already applies. A fifteen-year model holding 28% flat is asserting that a trend which removed a third of the rating in one year stops immediately.
The DSCR line that follows
Assumed energy revenue $17,520,000
Assumed opex $3,000,000
Debt service $10,755,556
DSCR at 28% ELCC (17,520 + 6,643 − 3,000) ÷ 10,756 = 1.97×
DSCR at 20% ELCC (17,520 + 4,745 − 3,000) ÷ 10,756 = 1.79×
DSCR at 7% ELCC (17,520 + 1,661 − 3,000) ÷ 10,756 = 1.50×
The project survives all three on these assumptions, which is the point of including it: accreditation risk is usually a return problem rather than a covenant problem, provided the structure was not sized on the capacity line. Where a structure was sized on it, the same table is a default schedule.
ℹ️ Note: The single most useful output is not a central case. It is the accreditation level at which the capacity revenue stops mattering to the credit — below which the lender's answer is the same either way. Publish that, and the negotiation about the central case becomes much shorter.
To build the accreditation decline series, the decomposition and the DSCR sensitivity, prompt Dezzmond with your technology, duration and market.
What Can a Developer Actually Do About It?
Four things, and none of them is arguing about the rating.
Add duration. This is the only response that directly buys back accreditation, and the ladder prices it explicitly: moving from four hours to eight hours takes the rating from 55% to 67%, and to ten hours takes it to 75%. Whether that is worth the incremental capital is an arithmetic question the model can answer, and it is one of the few places in this post where a project-level decision moves a market-level variable.
Hybridise. Pairing storage with solar attacks the problem at its root, because the reason solar's rating collapsed is that it cannot serve the evening ramp. A solar-plus-storage configuration can, and it accesses a materially different point on the accreditation curve than the solar alone does. It also, as the basis posts noted, addresses congestion and negative pricing at the same time — which is why hybrid configurations have become the default in the most saturated markets rather than an exotic option.
Site and configure for winter. If nearly ninety percent of loss-of-load risk migrates to winter, the resource characteristics that earn capacity revenue change with it. For wind that favours sites with strong winter production; for storage it favours duration over power; for solar it is largely unanswerable, which is the honest conclusion.
Do not bank it. The most reliable response is structural rather than technical: size the debt on energy revenue and treat capacity as equity upside. A project that survives the seven percent case in the sensitivity table is a project whose sponsor keeps the capacity revenue when it arrives and does not default when it does not. Every other response on this list is an attempt to hold a number that the market is designed to reduce.
What Do Sponsors and Lenders Actually Check?
- Is accreditation modelled as a separate input from the clearing price? If they are combined, the sensitivity cannot be run.
- Does the model decline the class rating over the tenor, and on what stated basis?
- For storage, what duration is assumed — the ladder runs 55% at four hours to 75% at ten.
- Is the project summer-peaking in a market whose risk is migrating to winter?
- How much of the technology class is in the queue behind this project? That is what sets the future rating.
- At what accreditation does the capacity revenue stop mattering to the credit?
- In MISO, has the indicative DLOL result been run rather than the historical capacity credit?
Frequently Asked Questions
What is ELCC?
Effective Load Carrying Capability — a measure of how much additional load a resource allows a system to serve at the same level of reliability. It values contribution to reliability, not energy produced.
What is solar's capacity accreditation in PJM?
For the 2028/2029 delivery year, 5% for fixed-tilt and 7% for tracking solar. A 200 MW tracking plant is accredited at 14 MW.
Why does storage accreditation depend on duration?
Because reliability events have length. A four-hour battery covers four hours of an event; system stress is migrating to winter, where events are longer. PJM's 2028/2029 ratings run 55% at four hours, 65% at six, 67% at eight and 75% at ten.
Why does accreditation keep falling?
Because ELCC is marginal. Each additional unit of a technology covers hours the previous units already covered, so its incremental reliability contribution declines — and the class rating falls for the whole fleet, not just new entrants.
What are Capacity Interconnection Rights?
The quantity of capacity a project is entitled to offer, established through the interconnection study process. Accreditation sets the percentage; CIRs set the base it applies to. A project with energy-only interconnection service may have none, and therefore no capacity revenue at all.
Can a good project get a better rating than a poor one?
Usually not. Class ratings apply to a technology class, not to individual assets, so a project with an excellent profile receives the same percentage as a mediocre one unless the market offers a resource-specific accreditation path.
What is MISO's DLOL methodology?
Direct Loss of Load accreditation, being introduced from Planning Year 2028–2029, which credits resources on availability during critical hours in MISO's probabilistic model, heavily weighting the top 65 at-risk hours in each season.
Closing: A Revenue Line That Falls Because the Technology Works
Most revenue risks in project finance are risks of failure. The plant underperforms, the counterparty defaults, the price falls, the contractor is late. Accreditation risk is the opposite, and that is what makes it hard to hold in mind.
Solar's class rating fell to seven percent because enough solar was built to solve the summer afternoon reliability problem. Wind's fell by a third because the risk hours moved. Four-hour storage is forecast to fall from fifty-seven percent to twenty because four-hour storage will have covered the four-hour events. In each case the technology did exactly what it was supposed to do, and the payment for doing it fell accordingly.
That is a coherent policy design — reliability payments should track reliability contribution, and a saturated resource contributes less at the margin. It is also a genuinely difficult thing to finance, because it means the revenue line declines for reasons entirely outside the project's control, with no contract allocating it and no counterparty to claim against.
The practical response is narrow but real: model the decline explicitly, size the debt without it, and know the level at which it stops mattering. A project that needs its capacity revenue to hold flat for fifteen years is relying on the rest of the market stopping construction.
The next post follows the duration ladder into the asset it was designed around: what a battery actually earns, product by product, and how the stack changed when ERCOT co-optimised its market in December 2025.
Sources: PJM — ELCC Class Ratings for the 2028/2029 Base Residual Auction · PJM — ELCC Class Ratings for the 2027/2028 Base Residual Auction · Ascend Analytics — PJM ELCC and Capacity Market Modeling · MISO — Wind and Solar Capacity Credit Report PY 2025-2026