What a renewable kilowatt displaces depends on the hour and the grid, and economists argue that subsidies paying the same rate for all of it are mispriced.

A megawatt-hour of wind power is a megawatt-hour of wind power. Its environmental value is not a property of the electricity, though — it is a property of whatever would have generated that electricity instead, and that varies by the hour and by the region.

A paper in The Energy Journal from the University of Colorado Boulder argues that this variation is large enough to make current renewable subsidies systematically mispriced, because they pay the same amount regardless of what a given unit of clean power displaces.

“You can have a technically identical megawatt-hour of energy from a wind farm or a solar plant,” explains Daniel Kaffine, a RASEI Fellow at CU Boulder. “But the value to society of that megawatt-hour, from an environmental standpoint, can be very different.”

The marginal plant is what counts

The mechanism is a feature of how grids dispatch. At any moment, operators are running the cheapest available generators and the last one switched on sets the price — and that marginal generator is the one a new unit of renewable output actually pushes off the system.

So the emissions a wind turbine avoids are the emissions of whichever plant was on the margin when it generated, not an average of everything on the grid. Displace coal and the avoided emissions are large; displace gas and they are roughly half that; displace another renewable and they are zero.

Time of day matters because the marginal plant changes through the day. Nighttime generation is more likely to displace coal, which runs steadily because it is slow and expensive to cycle. Midday generation more often displaces gas, or other renewables.

Location matters for the same reason at a larger scale. Wind in the coal-heavy upper Midwest probably offsets coal near population centres, which improves air quality as well as reducing carbon. Wind and solar in California mostly compete with existing renewables.

The California case

California supplies the sharpest illustration, and it is a situation that would have seemed absurd twenty years ago.

There, midday wind can force solar to be curtailed rather than displacing any fossil generation at all. Curtailment means an operator instructing a working solar farm to stop producing because there is nowhere for the power to go — which happens in California routinely, because so much solar capacity was built that midday supply exceeds what the grid can absorb.

A wind farm generating into that surplus produces clean electricity whose environmental benefit is approximately nothing, since the power it replaces was already clean.

“You don’t want to be subsidizing a wind generator that is just bumping off solar in the middle of the day,” explains Kaffine.

What the policies do instead

The instruments in use treat all renewable output as equivalent. Production tax credits pay per megawatt-hour generated; renewable portfolio standards require utilities to source a share of their power from renewables; renewable energy credits are traded as units of clean generation. None of them references what any particular unit displaced.

The authors’ two recommendations follow. First, measure marginal emissions in near-real time — which some upper Midwest grid operators are beginning to pilot, so the data problem is not insurmountable. Second, tie subsidies to environmental impact, in the way food labelling distinguishes between products that look alike.

There is a real argument on the other side, and the release does not present it. Simple, uniform subsidies are predictable, which is what allows projects to be financed over twenty-year horizons; a credit that varies by hour and location introduces revenue risk that raises the cost of capital, and could reduce total renewable build-out even while improving the environmental value of each unit. Flat policies have also been remarkably effective at driving deployment and cost reductions, which is the outcome they were designed for.

The counter-counterargument is that this is a problem of transition. A flat subsidy is well suited to a grid with little renewable capacity, where almost anything clean displaces something dirty, and poorly suited to one where midday solar is already being curtailed — which is the grid several US regions now have.

Pricing something with no market

Kaffine is candid about the underlying difficulty, which is valuation rather than measurement.

“We know clean air is valuable to people. We know that the environment is valuable to people,” he says. “But there’s no market for it, so we don’t have a price for it. How do you put a number on that?”

That is the question economics has approached for decades through damage estimates — the social cost of carbon, and separately the health costs of the sulphur dioxide, nitrogen oxides and particulates that coal plants emit near where people live. The air-quality component is the part that makes location matter so much, because carbon mixes globally while particulates harm whoever is downwind.

Those numbers are contested, and the policy proposal inherits the contest. A subsidy indexed to environmental impact requires an agreed figure for environmental impact, and agreeing one is a political process rather than an empirical one.

What the release does not contain

The paper’s title is not given, and the second author is identified only as “Fell” — no first name, role or affiliation. Neither funding sources nor conflicts of interest are stated.

There are also no numbers. Not one figure appears for how much the value of a megawatt-hour varies between the cases described: no emissions rates, no dollar values, no estimate of how much current subsidies over- or under-pay in any region. For a paper whose argument is that a quantity varies more than policy assumes, the magnitude of the variation is the result, and it is absent.

“At RASEI, the end game is always a better environment,” Kaffine says. “All the technical research is in service of making a cleaner environment. It is really cool to be able to trace these advances all the way through to the end impact.”

---

Notes

1. Kaffine, D., and Fell. The Energy Journal, 2026. DOI: 10.1177/01956574261462659. The release does not give the paper’s title and names the second author only by surname.
2. University of Colorado at Boulder. “Not all renewable energy is created equal.” EurekAlert!, 6 August 2026. https://www.eurekalert.org/news-releases/1139235.

Photo by Kervin Edward Lara on Pexels.com