The common figure for Malaysian electricity is 0.74 kg of CO₂ per kilowatt-hour. It is published by the Energy Commission, and it reproduces exactly from that body's own data. It applies only to Peninsular Malaysia.
Sabah is 0.539. Sarawak is 0.199. A data centre in Kuching costed at the Peninsular figure overstates its emissions by 3.7 times.
Two bodies, two methodologies for the same three regions
The Energy Commission publishes an average, defined as total emissions divided by net generation. Its own definition is explicit — "GEF is the generation weighted-average greenhouse gas (GHG) emission per unit of electricity generation". We re-derived all eighteen published cells; Peninsular 2024 is 110,555 Gg CO₂e over 149,301 GWh, which is 0.74048, printed as 0.740.
A second body, MGTC, publishes figures for the same three regions on a marginal basis, for carbon crediting. These diverge from the average, and the divergence changes direction by region:
| Region | Average basis | Marginal basis | Divergence |
|---|---|---|---|
| Peninsular | 0.740 | 0.585 | −23.7% |
| Sabah | 0.539 | 0.525 | −0.9% |
| Sarawak | 0.199 | 0.330 | +54.9% |
The mechanism is stated in MGTC's own methodology. The operating margin is the "generation weighted emissions averaged of all power plants serving the system excluding low cost/must-run", and those excluded plants "include hydro, geothermal, wind, low-cost biomass, nuclear and solar generation".
⚠️ The marginal basis excludes hydro by construction, and Sarawak is hydro-dominated. So Sarawak has the highest operating margin of the three regions, 0.9667, and the lowest average, 0.199. Same grid, same year, two official numbers 4.5 times apart, and neither is wrong for the job it was built for.
The test that settles what kind of number this is
The Philippines makes the point more cleanly than any argument could. Its National Grid Emission Factor gives the Luzon-Visayas grid, for 2020 to 2022, as:
0.7346 for wind and solar projects
0.7612 for other projectsOne grid. One period. Two values, and which one applies depends on what you are building.
A grid's carbon intensity should be a property of the grid. A figure that changes according to the project attached to it is not a measurement of electricity but a baseline, an estimate of what a new project displaces, which is a different question with a different answer. We reproduced all twelve published Philippine values from their own operating and build margins, at weightings of 0.75 for wind and solar and 0.50 for everything else. The document does not state those weightings anywhere.
Indonesia publishes 154 grid factors, all on the same marginal basis. Neither country publishes an average.
What this does not mean
These figures are not wrong; they are built for different purposes. A combined margin is the correct instrument for crediting what a new solar farm avoids, and the publishing bodies are following the CDM methodology as specified. MGTC even records the reuse in its own scare quotes, noting its baselines "were also used as 'national emission factor' to calculate the emissions associated with mitigation measures that saves electricity" — the quotation marks are theirs.
The Energy Commission, for its part, hides nothing. It states its basis in the first line, and it footnotes that its Sarawak column comes from a different publisher rather than passing it off as its own computation.
⚠️ And we did not establish why any of these choices were made. We are reporting what the documents publish and what the arithmetic does, and declining to attribute intent to any of it.
The source of the Sarawak figure
Sarawak's figure is not computed the way the other two are. The Energy Commission's tables carry six per-fuel breakdowns for Peninsular and six for Sabah, and none for Sarawak; the word "Sarawak" appears twice in the whole document, as a column header and in the sources note. The value is carried from Sarawak Energy Berhad's own annual report, where it is a corporate Scope 1 inventory figure — assured, GRI-framed, and a different kind of measurement from the other two columns.
Three columns, one table, and not the same kind of number in each. That is a comparability caveat rather than an error, and it is the sort of thing that vanishes when a figure is copied into a spreadsheet.
What to do with it
Never write "Malaysia = 0.74" without naming the grid. Ask which one, and use Sabah's or Sarawak's figure if that is where the meter is.
For Scope 2 reporting, use the average and not the margin. MGTC's own calculator gets this right — it instructs users to take the "Malaysia Grid Emission Factor depending on your state (Peninsular/Sabah/Sarawak)", pointing at the Energy Commission's average rather than at MGTC's own marginal figures.
In Indonesia and the Philippines there is no published average to reach for, so a Scope 2 calculation there imports a project-baseline construct whether you intend it or not. Say so in the footnote rather than presenting the number as a measurement.
And check the renewable-certificate caveat: the Energy Commission notes its factor includes renewable plants that may issue certificates, so it is not a residual-mix figure. Pairing it with purchased certificates counts the same clean electricity twice.
Where this comes from, and what will date it
The average-basis figures and the definition are from the Energy Commission's own "Grid Emission Factor (GEF) in Malaysia", published 25 November 2024, and its provisional 2022–2024 companion. The marginal figures, the methodology quotations and the regional weightings are from MGTC's 2017 CDM Electricity Baseline, a 49-page report. The Philippine values are from the Department of Energy's National Grid Emission Factor, and Indonesia's from the ESDM guidance. The Sarawak comparison is against Sarawak Energy Berhad's own annual and sustainability report.
⚠️ Two of the ten documents we read have no text layer at all — the Philippine factor extracts two bytes, and one Indonesian decree extracts thirteen. Both were read visually at high resolution instead, and any count we quote from those two rests on an optical layer we generated ourselves. We are flagging them as weaker evidence than the rest rather than presenting all ten as equivalent.
One further limit, in the other direction. Four different global-warming-potential vintages are in use across these documents, and three of the publishers state none at all. That sounds damning and mostly is not: swapping between them moves the Peninsular figure across a range of 0.04%. The reproducibility gap is real and the numerical consequence is negligible, and it would be easy to write the first half without the second. The gaps that matter are region, at 3.7 times, and basis, at −24% to +55%.
This dates slowly for the average figures, which are refreshed annually. The marginal baseline is already nine years old — its baseline year is 2017 — and MGTC's own site no longer links it. Re-derive before citing any of it.
View author profile → · Editorial policy
