Error in building VA 2020 two-region disaggregated model

[bl-young]

Error in buildTwoRegionUseModel(state, year, ioschema = 2012, iolevel = iolevel,  : 
  RoUS (of Virginia)'s commodity output summed from two-region (total) Use table doesn't equal to RoUS's commodity output.
Calls: source ... eval -> assembleTwoRegionIO -> buildTwoRegionUseModel
Execution halted

https://github.com/USEPA/stateior/actions/runs/4549476400/jobs/8021585025

[bl-young]

This is for disaggregated models, v0.3.0

[jvendries]

As mentioned in the stop message, this error occurs because there is a difference in the values calculated for the commodity output for the RoUS calculated using the US commodity totals and the RoUS calculated using the the two table structure (RoUS2RoUS + RoUS2SoI) for the non-domestic tables.

The specific sector in question is the Other commodity (last commodity in the table), where the value for the latter calculation is $2 million lower than the former. This is usually not a big enough value to create a > 1% difference threshold used as a check, but the value of the Other commodity is low enough in the first place that it results in a approximately a 1.1% difference. This is the only sector that fails this check for the 2020 disaggregated VA model.

Since this is only happening in the 2020 disaggregated version, I am still tracking where exactly in the disaggregation steps the difference is introduced to the Other sector, as this is not a sector that is treated any differently for this year than any other in the current test runs.

I'll post an update when I find the exact cause, but given the small difference I am inclined to suspect the error arises in the 2020 table due to a rounding error with smaller initial values in the VA tables for this particular year rather than an error with the disaggregation code per se.

[bl-young]

this issue is resolved following e1ed152
@jvendries please let me know if you agree and we can close this