Do we aggregate the private and government electric industries?

[WesIngwersen]

The electricity commodity is the primary product of 3 industries in the detailed make table:

221100, Electric power generation, transmission and distribution
S00101, Federal electric utilities
S00202, State and local government electric utilities

USEEIO leaves the industries as these 3. CEDA aggregates these industries into a single industry.

Should the industries be aggregated into a single industry?

Part of project https://github.com/orgs/cornerstone-data/projects/6/

RESOLVED: Consensus is to aggregate the 3 electricity industries.

[bl-young]

As it relates to the satellite tables:

In our GHG model, we have a single source for electricity generation emissions from the GHGI, which is simply attributed across all three sectors on the basis of gross output. In the CAP_HAP model, the federal and state utility sectors are not present - all facility level data are assigned via NAICS and so only 2211 is relevant. Similarly in the water model all electricity water use appears to be assigned to 2211*.

Presumably for the facility level datasets (CAP_HAP) we could likely identify specific facilities as belonging to 2211 or to federal or state and local (though I've not looked closely at this). But for the other datasets this would require some amount of attribution.

[WesIngwersen]

If the environmental data does not at least currently have the info to disaggregate them and if across various extensions we might have contrasting approaches, this suggests a need to aggregate them for consistency.

[jvendries]

In addition to this, and though not part of our immediate scope, the aggregation of all electricity sectors would likely be the first step for disaggregating electricity by 6-digit NAICS, which we have discussed as a potential item to pursue in the future.

[bl-young]

If we aggregate them, I expect some complications could arise as it relates to summary models and state models. At the BEA Summary level, GFE ("Federal government enterprises") includes S00101 and GSLE ("State and local government enterprises") includes S00202. In 2017, the two utilities sectors each made up about 16-18% of the total summary level sector.

[WesIngwersen]

This does impart a challenge on the state models. If we used Make and Use with electricity aggregated for this it will essentially require a redefinition, as BEA does it, to move a portion of output, inputs and uses from GFE and GSLE to 21.

[MoLi7]

In the multi-regional context, we observe that non-US IO tables and GHG inventories only have one electric power utility industry that aggregates all types and sources of generation. It will require a fair amount of attribution if we want to keep the government electric industries, plus data availability will be a big concern as there aren't high quality data about electricity generation by government facilities outside of the US.

[sangwonsuhvm]

Electricity is an homogenous commodity--there is no functional or quality difference in federally produced or other electricity sources. Furthermore, both federal and non-federal power generation share same grid network making it not possible to physically segregate producers and users. From the IO theory point of view, this is exactly the case of single commodity produced by multiple-producers. Given that direct emissions can't be distinguished among them either, aggregating them seems methodologically and practically preferable.

[bl-young]

In summary, the consensus is that aggregating these three industries simplifies the approach substantially with little impact on the end user due to the homogeneity of the electricity commodity. Additionally this aligns with most US environmental datasets (which do not distinguish federal vs private), aligns with non-US data, and facilitates future fuel-specific disaggregation.

While this could complicate the development of state models by reassigning portions of GFE and GSLE, this feels like a worthwhile tradeoff.