What is a dynamic baseline, and why does it make the avoided-emissions headline smaller?

Short answer. A dynamic baseline lets the incumbent change over time: grids decarbonise, incumbent technologies improve, the displaced mix shifts. A static baseline freezes today's incumbent for the solution's whole life and credits the solution with emissions that would have gone anyway. Carbbin models both sides year by year. It usually makes the headline smaller. It is what makes it hold.

Why the incumbent moves

A unit sold today runs for ten, twenty, thirty years. Over that time the grid that feeds it changes, the boiler or engine it replaces gets more efficient, rules phase some options out, and the market it sells into fills up with other clean products. If the comparison ignores all that, the solution takes credit for progress it did not cause.

The direction depends on the solution. One that saves grid electricity avoids less every year as the grid greens, and tends towards zero on a fully renewable grid. One that runs on grid electricity and replaces a fossil incumbent can avoid more over time. Both effects live in the shape of the curve, which is why an average hides them.

What the standards say

Project Frame (Guidelines 2024) separates the two cases. A dynamic baseline lets incumbent unit emissions change over time; a static one holds them constant, a simplification analysts sometimes accept "where change can't be reasonably projected" (Section 3, box, p. 21). Frame recommends calculating net unit impact year by year, from year one to the last year (Section 3, p. 22), and lists the usual drivers of change: grid emissions, technology, new effects, policy and market dynamics (Section 3, p. 23). Grid intensity is also one of the reasons unit impact differs by segment (Section 3, p. 16).

Frame's grid paper (2025) is more specific:

• For long-lived assets, today's grid intensity should not be applied beyond four to five years. Project it year by year and consider several scenarios (p. 4).
• Use or align with reputable forecasts, such as the IEA, the US EIA, NREL, the IPCC and national grid operators, and aim for the trend rather than exact values (p. 4).
• Between milestone years, assume a constant rate of change unless better projections exist, account for grid-specific policy, and run a sensitivity analysis (p. 4).
• Match the metric to the geography and to when the load runs: annual, seasonal, time-specific or marginal (pp. 3 to 5).
• The analyst always explains which method was used and why (p. 4). In the paper's words, "it is better to be roughly right than precisely wrong" (p. 6).

The ILCD Handbook reaches the same place from data quality. Studies supporting long-lived products may need conclusions valid for 20 to 30 years, which points to future scenarios for the use and end-of-life stages rather than present data (§6.8.4, p. 133). For processes more than five years in the future, fully time-representative forecast data should be used where possible (Provision 6.8.4 III, p. 135). Less representative data must not unfairly disfavour a competing product (§6.8.4, p. 134), and where a product is replaced during the comparison, the replacement should consider that a newer model may take its place (Provision 6.10 XI, p. 147).

ISO 14067 governs each footprint and has no counterfactual of its own. Two of its rules still shape the baseline: where emissions vary over time, data should use trend values where possible (6.3.6), and use-stage flows that occur more than ten years after a product enters use have their timing stated in the inventory (6.4.8).

The WBCSD Guidance on Avoided Emissions (v2.0) asks for trend scenarios, such as the IEA's, in forward-looking assessments (Step 4), and keeps the reference technology mix fixed for each year of sale while energy factors may evolve.

How Carbbin does it

• Carbbin applies each year's emission factors to both sides, for every year of the unit's service life. Same pathway for the solution and the incumbent, every year. That symmetry is what keeps the comparison honest.
• Carbbin uses a sourced pathway per market, from official or reputable scenarios for that country's grid, discloses it and bounds it.
• Carbbin runs the pathway over the full service life, not a ten-year cap. Where published data runs out, it holds the last credible value or applies a bounded decline, and says which.
• Carbbin models the incumbent as a mix that evolves where the displaced mix changes. As clean products take the market, the unit a new sale displaces is more often already clean, and only the remaining difference counts.
• Carbbin keeps the year-by-year curve, not only the average. A lifetime average is labelled as an average, never as the yearly rate.
• Carbbin tests the pathway. The range shows what happens if the grid moves faster or slower, so a reviewer can see how much rests on it.
• Carbbin does not rewrite earlier years once real sales are tracked. Each year's results stay as they were computed, so later updates never change past results. See potential, planned, realised.

"Too many assumptions?"

A fair question, often asked after 2022. A pathway is not a prediction of every shock; it is the sourced trend, with its bounds shown. And a static baseline is an assumption too: that the grid and the incumbent stand still for twenty years. That is the one assumption we know is wrong.

Common mistakes

• Today's grid factor for the whole life of a long-lived product.
• A dynamic pathway on one side only. If the solution's electricity gets cleaner, so does the incumbent's.
• A flat average that hides a declining curve, which matters most for solutions that save grid electricity.
• A pathway that stops at year ten for a product that runs for twenty.
• Full credit for displacing a fossil unit when, by then, the unit displaced is likely to be clean already.
• Mixing bases, such as an average-mix decline grafted onto a marginal factor. One method, both sides.
• No sensitivity on the pathway, so nobody can tell how much of the headline rests on it.
• An incumbent modelled on another place's grid, or several countries modelled with different trend methods in one comparison.
• The solution on a renewable supply contract, the incumbent on the grid. The gap then measures the contract, not the technology.
• An incumbent still credited after its legal phase-out date.

Sources

• Project Frame. Project Frame Methodology: Evaluating Greenhouse Gas Impact for Early-Stage Investments. December 2024. https://projectframe.how/s/2024-Project-Frame-Methodology.pdf
• Project Frame. Quantifying GHG Emissions Intensity of Electricity Grids: Selecting the Right Metric for Net Unit Impact. September 2025. https://projectframe.how/additional-resources-blog/ghg-emissions-grid-intensity
• European Commission, JRC. ILCD Handbook: General guide for Life Cycle Assessment, Detailed guidance. First edition, March 2010. §6.8.4 and Provisions 6.8.4 and 6.10.
• ISO 14067:2018. Greenhouse gases: Carbon footprint of products. Requirements and guidelines for quantification. Clauses 6.3.6 and 6.4.8.
• WBCSD. Guidance on Avoided Emissions, version 2.0. 2025.