Long-Run vs Short-Run Marginal

Marginal emissions is the right number for policy calculations. But "marginal" hides a question: marginal over what time horizon? The answer changes the number dramatically, and gets the answer wrong in opposite directions depending on which you pick.

Short-run marginal emissions measure which plant ramps up tonight if demand increases; long-run marginal emissions measure which plant gets built over the next decade if demand increases persistently. Short-run marginal is almost always a gas plant (the flexible swing producer). Long-run marginal, in most markets today, is likely to be solar-plus-storage or wind, because those are the resources at the margin of investment decisions.

The counterintuitive implication: the same incremental load can be high-carbon (short-run) or near-zero-carbon (long-run) depending on which question you're actually trying to answer.

A city is deciding whether to mandate heat pump installation in commercial buildings over the next 15 years. The policy will add persistent, predictable load to the grid.

Should the analysis use short-run or long-run marginal emissions?

Long-run marginal. The city's heat pump load won't be met by tonight's gas dispatch; it will be met by whatever capacity gets built to serve it. In 2024, the marginal investment in most U.S. markets is solar or wind. Using short-run marginal emissions would make the policy look dirtier than it is.

Contrast with a real-time demand response program that shifts loads hour by hour. That decision genuinely affects tonight's dispatch, so short-run marginal applies.

Carbon accounting that conflates the two time horizons produces systematically biased policy recommendations. Electrification programs evaluated against short-run marginal look worse than they are; programs evaluated against long-run marginal may look cleaner than near-term reality. The correct choice depends on the actual causal pathway: is this decision affecting dispatch or investment?