Building the WiNDC National Data
August 15, 2025
Today I am going to discuss the build process for the WiNDC National data. I will demonstrate taking the data from the raw Supply/Use tables and transforming it into the final National data product.
As a quick follow up to the post on August 1, I stated
that we “we make values from use positive and
supply negative. Much like Ben Franklin declaring the
charge on an electron to be negative, I have made the wrong decision.
However, at this stage of development it easy to change our convention,
which I have. Credit goes to Mei Yuan who pointed out this issue during
the WiNDC Weekly this past week.
This post will be using terminology from WiNDCNational.jl.
Raw Data to Pre Calibrated Data
The only necessary data for the national model is the BEA Supply and Use tables. These tables have the following structure:
Note that the Other_Final_Demand region is a collection
of multiple columns. However, in the national model we don’t distinguish
between the values so they are all treated as a single category. The
data does not get aggregated together, retrieving that parameter returns
all the raw data.
Coming from the BEA, this data is not suitable for use in our CGE models. In the following sections we will outline the steps taken to transform the raw data into a format suitable for our models.
Flip Sign of Use
The first step to flip the sign of all the values in the Use table. We consider inputs to be negative.
Reverse Negative Flows in Intermediate Tables
We want all values in the Intermediate tables to have
the same sign as their flow. For example, in the 2023 Use table the
sector 111CA (Farms) and commodity Used
(Scrap, Used and secondhand goods) has a value of 18. We
expect all values in the Use table to be negative. Note, this is after
flipping the sign from the previous step. In the Use table that value is
-18.
To solve this issue we reverse the flow of good with the wrong sign
in Intermediate_Demand and
Intermediate_Supply. The 18 moves from Use to
be 18 in Supply in the same commodity/sector.
Redistribute
CIF/FOB Adjustments
The CIF/FOB column is used to adjust
Transport margins and Imports. The
CIF/FOB values get added to Imports values on
insurance commodities and Transport on non-insurance
commodities. In the Summary-level table their is only a single insurance
commodity, 524 (Insurance carriers and related activities).
In the detailed table there are three:
524113(Direct life insurance carriers)5241XX(Insurance carriers, except direct life)524200(Insurance agencies, brokerages, and related activities)
The CIF/FOB column is then removed from our data.
Disaggregation of
Trade and Transport
The two marginal columns Trade and
Transport are transformed into two parameters,
Margin_Demand and Margin_Supply. The positive
values from Trade and Transport become
Margin_Demand and the negative values become
Margin_Supply. We do not adjust the signs of these values,
Margin_Supply acts like an input which means it should be
negative.
We also create a new set margin which includes the NAICS
codes for Trade and Transport. This set is the
column domain of both Margin_Demand and
Margin_Supply.
Personal_Consumption
and Household_Supply
The Personal_Consumption column has a mix of positive
and negative values. This could should only contain negative values, as
it is on the input side. The Household_Supply column is
created from the positive values.
Sector_Subsidy and
Subsidy
The Sector_Subsidy is a new addition to the Supply/Use
framework. It was introduced post-Covid to account for the large
subsidies provided to various sectors of the economy. Direct from the
BEA, this row is positive and it should be negative. That means
in our final data, this row should be positive, so we re-flip the
sign.
This is in contrast to the Subsidy column which, direct
from the BEA, is negative. Which is the correct sign so we make no
adjustments.
The Final Data Structure
The final WiNDC National parameters are given by:
These are the actual parameters in the WiNDC National dataset. This
means if you want to extract Export, you can run
table(summary, :Export)and get the corresponding values from the dataset.