How do we make emission factors in purchaser value?

[WesIngwersen]

Emission factors can be prepared in purchaser value, which includes the value of wholesale/retail and distribution (aka margins in IO language). Factors in this valuation are more appropriate for users that want to multiply spend data by emission factors to get final emissions.

In USEEIO, emission factors are first computed using the model's valuation, which is in producer price, and the models scope, which as defined by a producer price model is "cradle to gate". The PRO factors are converted into PUR factorse using commodity-specific margin rations of $ margin per $ PRO, which gives the Supply Chain Emission Factors (SEF). Also the impact of margins is calculated based on the wholesale, retail and distribution portions of that margin but specific margin commodities within that are not used. These are estimated in PUR price and called "Margin Emission Factors" (MEFs). These are added together to get the final factors which represent a "Cradle-to-shelf" for commodities purchased at a retail store, or in the case of items shipped from a retailer to the final user, "Cradle-to-user".

$SEF (PUR) + MEF (PUR) = SEF with margins (PUR)$

[WesIngwersen]

Even assuming we start with a PRO model the data requirements for this transformation suggest that it be included in milestone 1

[WesIngwersen]

Adding more clarity to the USEEIO approach, there are two steps:

1. The PRO -> PUR price transformation
   In USEEIO, the Detailed Margins table, which only exists for benchmark years, is used as the first source. The table includes the commodity, the purchaser, the producer value of the commodity, a wholesale, a retail, and a transportation value for every non-zero transaction in the Use table. Records for margins for export, import and change in inventories are removed, some commodities are remove and other adjustments are made. This is done in getMarginsTable()
   https://github.com/cornerstone-data/useeior/blob/90f4d7a022a3ecb2043535c30b473cd1068078a8/R/LoadMargins.R#L6-L67

The margins table is then converted into the Phi matrix in calculateProducerbyPurchaserPriceRatio()
https://github.com/cornerstone-data/useeior/blob/90f4d7a022a3ecb2043535c30b473cd1068078a8/R/AdjustPrice.R#L47-L66
In this Phi preparation, inflation/price years adjustments are made to the commodity producer price based on the commodity Rho value and to the margins sectors based on a margins sector Rho value.

Thus the application of Phi ratios used for this conversion are both commodity and year-specific, although the ratios themselves are based on those in the benchmark year.

2. The estimate of the impacts of the margins
   This is done in useeior in calculateMarginSectorImpacts()
   https://github.com/cornerstone-data/useeior/blob/90f4d7a022a3ecb2043535c30b473cd1068078a8/R/CalculationFunctions.R#L487-L524
   Margins coefficients which are margin:PRO ratio are calculated for each commodity.
   For each type of margin, e.g. transportation, this ratio is allocated to specific transportation sectors based on their total commodity output (acknowledging that transportation type is not commodity-specific)
   Those quantities representing inputs of margins put commodity are put into an A matrix like format (commodity x commodity) where the rows are the margin sectors and columns are commodities using them) and that gets multiplied by existing M and N matrices. Note the resulting matrices called "M_margin" and "N_margin" are still in PRO value (which is counter-intuitive).

Note that in application like here, then both the values taken from N and N_margin are price adjusted using Phi, and then can be reported together where they represent the SEF and MEF, respectively, in PUR value.
Example of application is here in the Supply Chain factors preparation:
https://github.com/cornerstone-data/supply-chain-factors/blob/1ecb5ff6a2464d52ce0b37a5f1c859a25b0cb069/CalculateEmissionFactors.Rmd#L20-L38

[MoLi7]

In CEDA, emission factors (EFs) are generated by default in producer price, too. The PRO EFs are converted to PUR EFs by multiplying with the producer-to-purchase ratios derived from the commodity-specific Margins table by BEA.

[WesIngwersen]

Clarifying after discussion with @MoLi7 , that in CEDA, there is:

1. PRO -> PUR price transformation, but there is no filtering of the BEA Margins table data as described for useeior above.

2. There is no equivalent estimation of the environment impact of the margins, due to uncertainty in the data.