Imports of Other and Used gets dropped

[WesIngwersen]

USEEIO/import_factors_exio/useeio_imports_script.py

Lines 208 to 220 in c4eb170

	t_df = get_tiva_data(year)
	corr = (pd.read_csv(conPath / 'bea_imports_corr.csv',
	usecols=['BEA Imports', 'BEA Summary'])
	.drop_duplicates())
	# ^^ requires mapping of import codes to summary codes. These codes are
	# between detail and summary.
	t_c = (t_df
	.reset_index()
	.rename(columns={'IOCode': 'BEA Imports'})
	.merge(corr, on='BEA Imports', how='left', validate='one_to_many')
	.drop(columns='BEA Imports')
	.groupby('BEA Summary').agg('sum'))

There are significant imports of Other goods and some of Used goods (see rows from t_df from 2012 these are in million current USD)

IOCode	CA	CN	EU	JP	MX	APAC	ROW
Used	2907	325	1456	179	713	360	578
Other	7766	3381	55516	12738	5021	15549	20653

but they get dropped after this merge because they are not in the concordance.

These would account for substantial embodied emissions and need to be handled.

[WesIngwersen]

A test of the quantity of imports before and after merging in this step could help identify and prevent such data loss

[bl-young]

Review those exiobase sectors that don't match at all - those are likely to be in this "other" category

[bl-young]

USED data was getting dropped because the sector mapped to USED from exiobase has no exports to the US (in exiobase). this is fixed with bf995e5

[WesIngwersen]

The BEA Concepts and Methods in IO (the IO Manual) has a good section on defining non-comparable imports and the source data.
These are mostly services consumed by US residents abroad. Foreign travel is a big one.
This needs to be address in the longer term, as they these services do have embodied flows.