Is the core schema for the GHG model based on BEA or NAICS schema?

[bl-young]

Flowsa uses NAICS as the schema (which gets mapped to BEA sectors in useeior) while CEDA directly aligns with BEA. The flowsa schema uses between NAICS 3 - NAICS 6 depending on the mapping required for BEA. That is, we only map to 3 or 4 digits when that is sufficient for a single BEA code (e.g., 2213 is sufficient for 221300 instead of attributing all the way to 221310 221320 and 221330).

What should be the target schema in the new model?

Resolved: A NAICS-based schema, hybridized with BEA sectors for select cases.

[bl-young]

While theoretically flowsa was designed to allow for alternate schemas, in practice it became challenging to enable the use of BEA schema directly. We have utilized some custom sectors to facilitate better alignment with NAICS, e.g., by adding quasi-NAICS for F01000. Still, in some specific edge cases, where NAICS to BEA mapping is not one:many or many:many, the conversion in useeior is not clear.
Specifically this manifests in the construction sector (23*) where NAICS and BEA are not well aligned, and this proves problematic when using something like the Use table for attribution to construction sectors. Gov't sectors (primarily 92* in NAICS) and housing (531*) can also be problematic.

[bl-young]

Further expansion of the custom NAICS list, e.g., to add the BEA construction sectors, would likely help alleviate the problem of BEA -> NAICS -> BEA that can muddle the data in these edge cases.

[MoLi7]

I like the expansion idea to address the challenges with 23* and similar cases, assuming the NAICS 23* codes will be removed from the SAM model to avoid duplication.

[bl-young]

Correct, we would not report both 236*, 237*, 238* (NAICS) and also the BEA 230* and 233*. Flagging this existing mapping file in useeior for reference.

[bl-young]

A NAICS-based approach does provide a bit of future proofing for future disaggregation if the emissions are already assigned by NAICS and we are as explicit as possible when doing that mapping (i.e., assigning to 5 or 6 digit if possible even if only 4 digit is the unique BEA code)

[WesIngwersen]

NAICS is more authoritative and more true to some industry environmental data. In the longer term is a better solution. However where their are truly unique identifiers in the BEA system as is done for construction, real estate and government as well as for final demand entities, we need to map directly to BEA and not let a forced NAICS intermediate mapping cause the attribution to be less accurate.

[MoLi7]

Cornerstone team's discussion favors a NAICS-based, BEA-compatible schema for GHG model. Specifically, we will create a hybrid schema that is primarily NAICS-based but includes additional BEA elements, which builds off of FLOWSA already has. For those NAICS codes like 23* and 92*, we will add BEA sectors to the schema.

We believe this approach avoids complex mapping while maintaining compatibility with both systems.