The GHG Protocol is the most widely-used framework for corporate carbon accounting. It is built into SECR, ESOS, CSRD, SEC disclosure proposals and most national equivalents. It is also a linear standard, designed around a take-make-use-dispose model of hardware. That made sense when it was written. It does not for an industry trying to migrate to circular practices, because the math actively disincentivises the migration.

Our recent paper, presented at the International Life Cycle Management conference (LCM 2023), sets out the problem and proposes a fix.

What the GHG Protocol gets wrong about second life

The current accounting treatment of embodied carbon in IT hardware works like this: when a server is manufactured, the full embodied carbon footprint of its production is attributed to the first owner. When the first owner disposes of the server, that embodied carbon is "spent" in their books. If the server is then refurbished and sold into a second life, the second owner inherits a server with zero accounted embodied carbon, because all of it was attributed to the first owner at manufacture.

This produces a perverse incentive. The first owner has no carbon-accounting reason to maximise the operational life of the server, because the embodied carbon hit happened at purchase regardless. The first owner has no carbon-accounting reason to choose disposal channels that route to refurbishment rather than destruction, because both look identical on the accounting line. The second owner gets the carbon benefit only of operational savings, not of the avoided manufacture. The accounting wraps the decision in a frame that rewards exactly the behaviour the circular economy is trying to discourage.

The proposal: amortize, then transfer

The fix we propose is structurally similar to how accountants already handle depreciation of capital assets. Embodied carbon should be amortized over a fixed asset lifetime, with a portion written down each year of operation. For servers, the appropriate amortization period is six years or longer, in line with measured operational life when properly maintained.

When a server is sold into the secondary market before the amortization period ends, the residual embodied carbon should transfer to the second owner. The first owner reports their actual fraction of consumed embodied carbon, not the full amount. The second owner inherits the rest and reports it as their own embodied carbon footprint going forward.

What the fix changes

Three incentives flip. All three flip toward circularity.

First, the first owner has a clear carbon-accounting reason to maximise operational life. Each additional year of use reduces the amortized carbon cost reported in that year.

Second, the first owner has a clear carbon-accounting reason to choose disposal channels that route to refurbishment. Selling the server into second life transfers the remaining embodied carbon away from the first owner's books, where destruction does not.

Third, the second owner has a credible incentive to choose refurbished hardware over new. Refurbished hardware enters their books with less embodied carbon than a newly-manufactured equivalent, even after the transfer.

These are the three behaviours every circular economy framework asks for. Linear carbon accounting works against all three. Amortization with transfer works with all three.

Where it fits with the broader picture

The proposal is not a wholesale rewrite of the GHG Protocol. It is a methodological adjustment within the existing scope 1, 2 and 3 framework, sitting in scope 3 (the embodied portion). It is consistent with the standard practice of amortizing capital assets in financial accounting, which makes it familiar territory for the CFO function increasingly involved in carbon reporting.

The GHG Protocol explicitly encourages multi-stakeholder processes to develop additional guidance. Carbon amortization with end-of-first-life transfer is exactly the kind of guidance the protocol's evolution should incorporate next, especially as circular-economy regulation (the EU Ecodesign Directive, Circular Economy Action Plan, equivalent regimes elsewhere) starts asking organisations for hardware-cycle reporting the current methodology cannot honestly produce.

Read the paper

The full research is available open access on ResearchGate.