Add the other three identification parts of Schedule R

md_ingest_relations() reads Schedule R Part II - related tax-exempt organisations. The form has three more identification parts, and all three name a counterparty with an EIN:

Usage

md_ingest_schedr_parts(
  con = NULL,
  which = c("disregarded", "partnership", "corporation"),
  version = "v2_2"
)

Arguments

con

A connection from md_db().

which

One or more of "disregarded", "partnership", "corporation".

version

Pinned version of the three Schedule R part sources, e.g. "v3_1".

Value

Row counts per part, invisibly.

Details

Part I

disregarded entities - efile_disregarded

Part III

related organisations taxable as a partnership - efile_related_ptr

Part IV

related organisations taxable as a corporation or trust - efile_related_corp

These counterparties are mostly not nonprofits, and the rate is the point

Part II counterparties are exempt organisations by definition, so their EINs resolve well. These do not. Measured on Part I across 2009-2024:

  rows                              2,000,291
  with a counterparty EIN             317,232
  distinct counterparty EINs           35,306
  resolving to a known `entity`         4,763  (13.5

13.5% is close to the 14.8% seen on 527 Schedule R, and for the same reason: a disregarded entity is usually a single-member LLC with no exempt status of its own, and a taxable subsidiary is a C-corp. They are outside the BMF universe.

Rows are still written with ein_b populated, because unlike the 527 file these sources give the EIN - an EIN that is real but unknown to matchdb is a different thing from an EIN that was never disclosed. A consumer joining ein_b to entity should expect roughly one in seven to land, and the per-source rate is reported on every run.

Part I is a different relation from Parts III and IV

See the §4a vocabulary. Parts III and IV describe organisations the filer controls, owns a share of, or shares governance with - related_entity. Part I describes an entity with no separate tax existence at all, whose activity is reported on the filer's own return. That is a tighter tie than any disclosure on the rest of the form, so it gets its own value, disregarded_entity.

The part is additionally carried in evidence (schedule_r_part_i_disregarded, ..._part_iii_partnership, ..._part_iv_corporation), so the taxable/exempt distinction inside related_entity survives without needing its own enum value.

A deferred question about disregarded entities

Because a disregarded entity has no separate tax existence, its name is arguably a name the parent operates under - closer to a dba than to a relation at all. That would be a name row on the parent's EIN.

It is deliberately not done here. §10 forbids casually aliasing one organisation's name onto another, and the difference between "disregarded for tax" and "the same organisation for matching" has not been measured. Recording the relation keeps the option open; writing the alias would not be reversible.

An ingest hazard, and it is not the one documented for 527

Pre-2013 files carry doubled quotes inside unquoted fields (PLEASE SEE SCHEDULE ""O"".), which is malformed CSV and aborts a strict parse partway through 2010. quote='"', escape='"', ignore_errors=TRUE reads them. Early years are also much wider - Part I is 96 columns in 2012 against 34 in 2013 - so union_by_name=TRUE is required, and most rows are blank padding rather than data.