Skip to contents

List the reporting areas whose data is folded into another area code

Source: R/polity_folds.R
folded_reporting_areas.Rd

A FAOSTAT reporting area is folded when polity_area_crosswalk gives it a polity_area_code that is not its own area_code. Every row the area reports is then summed into that other bucket, so the area disappears from WHEP output without a single row being dropped or left unresolved. This lists those areas, because the coverage reports cannot: a fold resolves perfectly well, only to a territory that did not report the data.

Three kinds exist, and they are not equally defensible:

  • "fabio_rest_of_world": FABIO collapses the area into its single Rest-of-World row (polity_area_code 999, ROW-1850-2025) because its own region list does not enumerate the area either. 57 areas, all flagged cbs FALSE in regions_full. Several still report substantial data of their own – Reunion, Guadeloupe, Palestine, the Faroe Islands – which is attributed to Rest of World.

  • "cbs_reporter_folded": the area is flagged cbs TRUE, so regions_full says it has a commodity balance sheet of its own, and it is folded into 999 anyway. Four areas: 153 New Caledonia, 154 North Macedonia, 209 Eswatini and 212 Syria, the last being the largest single contributor to the fold. FABIO does not fold these: its published region list enumerates all four as regions in their own right (see the section below), so this fold is WHEP's, not a FABIO convention, and the "fabio" label the other 57 carry does not apply.

  • "successor_state": the area is summed into the bucket of the state that succeeded it, which is a deliberate territorial identity rather than a loss: FAOSTAT area 62 "Ethiopia PDR" into 238 Ethiopia, and areas 276 Sudan and 277 South Sudan into 206 Sudan (former).

Whether to lift the Rest-of-World fold is an open decision recorded in issue 419; this function only makes the current state visible and changes nothing. A build also warns, naming the areas and the row counts it actually folded.

Usage

folded_reporting_areas(crosswalk = NULL)

Arguments

crosswalk

Crosswalk to inspect. Defaults to polity_area_crosswalk.

Value

A tibble with one row per folded reporting area, ordered by area_code:

  • area_code: The reporting area whose data is folded away.

  • area_name, area_iso3c: Its name and ISO3-like code.

  • polity_area_code: The bucket its rows are summed into.

  • polity_code, polity_name: The polity the fold attributes them to.

  • fold_kind: "fabio_rest_of_world", "cbs_reporter_folded" or "successor_state".

What FABIO's own region list says

FABIO (Bruckner et al. 2019) publishes the region list it uses, and it contains all four "cbs_reporter_folded" areas as regions of their own:

  • io_codes.csv of the FABIO v1.1 release (Zenodo record 2577067, the file inst/scripts/compare_fabio.R already downloads) enumerates 192 areas x 125 commodities. Areas 153, 154, 209 and 212 each have their own 125-row block, distinct from area 999 RoW.

  • The FABIO source repository (https://github.com/fineprint-global/fabio) folds an area into Rest of World exactly when it is absent from inst/regions_full.csv with current == TRUE. All four carry current TRUE there, and the 192 codes that file flags cbs TRUE are precisely the 192 areas of io_codes.csv.

So fabio_code == 999 for these four is a statement WHEP makes, not one FABIO makes. Correcting it in regions_full would move published values, because polity_area_code is derived from fabio_code, so the contradiction is left standing and reported here instead (issue 556).

The Rest-of-World fold is no longer applied

WHEP models every reporting member of bucket 999 in its own right. FABIO's 192-country layout is a methodology this package compares against, not a constraint on which territories it represents, and the choice of country set is WHEP's to make (issue 459).

That matters because the fold was never doing what its name suggests. Of the 61 members, only about a third report anything at all; the rest contribute no rows and folding them is arithmetically a no-op. Everything the bucket actually carried came from the members that DO file returns – Syria, Eswatini, North Macedonia, New Caledonia, the Faroe Islands, Palestine, Greenland and the like – and folding them discarded whose data it was. So promotion is self-limiting: an area with no rows is unaffected either way.

Bucket 999 survives as a genuine residual for the territories that report nothing. Measured on a full-range get_wide_cbs() (1850-2023), promotion takes the published area count from 195 to 216 and moves global totals by at most 0.99% (stock_addition), with every other column inside 0.4%.

options(whep.unfold_rest_of_world = "none") restores the fold, which is what reproducing a number published before this change requires. Because that no longer matches the published series, every read of the crosswalk warns while it is set. "cbs_reporters" re-folds all but the four "cbs_reporter_folded" areas and warns for the same reason. The "successor_state" folds are never lifted by any mode, since those are territorial identities rather than a FABIO convention.

An earlier measurement recorded in issue 419 reported this change at up to 13.7x on feed; that comparison predates the dcast() duplicate-key fix in .select_best_source() (issue 425) and does not reproduce.

References

Bruckner, M., Wood, R., Moran, D., Kuschnig, N., Wieland, H., Maus, V., Borner, J. (2019). FABIO - The Construction of the Food and Agriculture Input-Output Model. Environmental Science & Technology 53(19), 11302-11312. doi:10.1021/acs.est.9b03554

Examples

folded <- folded_reporting_areas()
nrow(folded)
#> [1] 10
head(folded[folded$fold_kind == "successor_state", ], 4)
#> # A tibble: 4 × 7
#>   area_code area_name    area_iso3c polity_area_code polity_code   polity_name  
#>       <int> <chr>        <chr>                 <int> <chr>         <chr>        
#> 1        62 Ethiopia PDR ETH                     238 ETH-1800-1889 Ethiopia (to…
#> 2        62 Ethiopia PDR ETH                     238 ETH-1889-1897 Ethiopia (18…
#> 3        62 Ethiopia PDR ETH                     238 ETH-1897-1902 Ethiopia (18…
#> 4        62 Ethiopia PDR ETH                     238 ETH-1902-1907 Ethiopia (19…
#> # ℹ 1 more variable: fold_kind <chr>
folded[folded$fold_kind == "cbs_reporter_folded", ]
#> # A tibble: 0 × 7
#> # ℹ 7 variables: area_code <int>, area_name <chr>, area_iso3c <chr>,
#> #   polity_area_code <int>, polity_code <chr>, polity_name <chr>,
#> #   fold_kind <chr>