Dropping factor where CO2 > 100 for entire country/commodity? about useeio HOT 10 CLOSED

I've compared the aggregate national factors when outliers are not excluded to the original approach. Where there is a difference greater than 5% only occurs in 4 sectors: 22, 212, 213, 562. The majority are in 22 where EFs can be several times greater 3-7x.

from useeio.

For 22, it is mostly due to changes in APAC and MX. I"m trying to track down why there are imports from APAC for utilities.

^^ @Jnamovich can you confirm where this BEA service mappings comes from

from useeio.

For 22, it is mostly due to changes in APAC and MX. I"m trying to track down why there are imports from APAC for utilities.

USEEIO/import_factors_exio/API/BEA_API_Mappings.csv

Line 55 in 48656ff

"Technical, trade-related, and other business services",TechTradeRelatedOth,221100

^^ @Jnamovich can you confirm where this BEA service mappings comes from

I could potentially see this being used by U.S. military bases in the Indo-Pacific. https://crsreports.congress.gov/product/pdf/R/R47589

from useeio.

For 22, it is mostly due to changes in APAC and MX. I"m trying to track down why there are imports from APAC for utilities.

USEEIO/import_factors_exio/API/BEA_API_Mappings.csv

Line 55 in 48656ff

"Technical, trade-related, and other business services",TechTradeRelatedOth,221100

^^ @Jnamovich can you confirm where this BEA service mappings comes from

Those were manually developed mappings in the absence of a robust concordance between BEA services and detail codes. Since the 22 detail sectors didn't map to any Census imports categories, I felt that I needed to map it to one of the service categories and this seemed to be the most appropriate of our options. That said, there is no reason to believe that this sector maps to 'Technical, trade-related, and other business services'. I had a look into the more detailed levels of the service codes and there is no indication that utilities is captured by that service category. Perhaps we should take this lack of explicit representation in the service or goods categories as an indicator that no imports occur in 22?

from useeio.

For 22, it is mostly due to changes in APAC and MX. I"m trying to track down why there are imports from APAC for utilities.

USEEIO/import_factors_exio/API/BEA_API_Mappings.csv

Line 55 in 48656ff

"Technical, trade-related, and other business services",TechTradeRelatedOth,221100

^^ @Jnamovich can you confirm where this BEA service mappings comes from

Those were manually developed mappings in the absence of a robust concordance between BEA services and detail codes. Since the 22 detail sectors didn't map to any Census imports categories, I felt that I needed to map it to one of the service categories and this seemed to be the most appropriate of our options. That said, there is no reason to believe that this sector maps to 'Technical, trade-related, and other business services'. I had a look into the more detailed levels of the service codes and there is no indication that utilities is captured by that service category. Perhaps we should take this lack of explicit representation in the service or goods categories as an indicator that no imports occur in 22?

I agree this is not a good mapping. Please remove it.

from useeio.

Yes we import electricity from CA and MX and this is reflected in TIVA import matrices

Data don't quite match up here, but TIVA import being greatest from CA is accurate.
Its appears the TIVA folks just spread a residual equally across all other non-CA TiVA regions.

At this point i don't suggest overwriting the import matrices to reflect the EIA data, although it could be considered later.

For now, the only thing significant in our calculations here would be the CA emission factor.

So i do recommend that we make sure that we're getting a CA electricity factor from EXIOBASE.

from useeio.

Yes @Jnamovich were wondering about this today - neither the census nor BEA data should really capture electricity so not sure where TiVA got their data from. Ultimately the mapping above should only impact the splits within regions (APAC, EU, etc.). We just need to confirm that when we drop that mapping it doesn't zero out this share above of < 3%

from useeio.

Because TiVA is only used in cases where we don't have imports data from BEA/census, this is now resolved. We use tiva for electricity which means the factor is >97% canada

from useeio.

@WesIngwersen @Jnamovich, we discussed pulling in the EIA data for electricity as noted in #24 (comment). However after I made the swap for electricity to aggregate sectors (in this case, fuel type) using industry output from exiobase instead of exports (non-existent) (in e800431) the factors look much better. So while using the EIA data might change things slightly, I think its less urgent now.

from useeio.

noting that @Jnamovich updated to use the EIA electricity data in fbd26dd

from useeio.