Who owns the lenders

A missing parent is not missing data

Read as a match rate, the parent edge alone looks thin: 78 of 300 top lenders, 26.0%. Adding the filed exceptions inverts the reading, and the two never overlap: across all 10 population and question combinations below, the largest number of entities carrying both a parent edge and an exception for the same question is 0, which is why the counts can be added at all. Run over the whole file the same way, 1,941,016 LEI records are live and not branches, which is the population the reporting rule binds; 93,540 of them report an ultimate parent and 1,847,346 filed an exception, leaving 130 that answer neither way, 0.0067%. Widen the frame to every record GLEIF has ever published, including the 1,441,783 whose registration is no longer issued, and the accounted share falls to 96.26%. That fall is not a reporting failure by anyone still reporting: 125,143 of the 126,562 records that answer neither way, 98.9% of them, sit on a registration that is no longer issued.

Source: GLEIF, LEI golden copy files | GLEIF, State Transition and Validation Rules v2.8.5 The reporting duty is GLEIF's: with the exception of branches, all legal entities shall report relationship data for both direct and ultimate accounting consolidating parent, and each relationship shall be reported by either a relationship record or a reporting exception record. Methodology

Both questions, every population

Source: GLEIF, LEI golden copy files "Live LEIs" are records whose registration status is ISSUED; the row above it removes the 1,323 of them that are branches, which GLEIF's rules exempt from reporting a parent at all. Entities are counted once per question, and the build refuses to write its output if the parent and exception counts do not sum to the accounted count. Methodology

Weighted by what the lenders actually lent

Counting entities gives the same weight to the largest lender in the curated table, which originated $139.7B in 2024, and to the smallest, which originated $63.5M. The table carries that origination volume, and all 300 of its 300 lenders join to this mapping on the LEI, so the same accounting can be weighted by money. The lenders that name an ultimate parent originated $541.6B of the $1.44T total, 37.69%; those that filed an exception originated $893.8B. The lenders with no answer of either kind account for $1.6B, 0.110% of the volume, so 99.890% of what these lenders lent was originated by an entity whose ownership status is on the record.

The reasons are the finding

An exception is not a refusal, it is a categorised answer. The 6,258,523 exceptions filed worldwide by 3,142,174 entities use 10 distinct reason codes, and the most common of them is Natural persons (2,297,728 filings, 36.7% of all of them). Among the top lenders the reason given most often is Parent does not consolidate, on 81 of the 220 exceptions filed. The reason attached to the most origination volume is a different one: No known controlling person, on 48 lenders originating $260.6B, against $251.0B for the reason given most often. Counting lenders and weighting them by lending do not rank these reasons the same way.

Source: GLEIF, Reporting Exceptions v2.1 format Only the first reason on each exception is counted. GLEIF allows up to five, and 13,650 of the 6,258,523 exceptions worldwide carry a second, 0.22% of them. Methodology

Source: GLEIF, Reporting Exceptions v2.1 format | CFPB/FFIEC, HMDA Loan/Application Register Volume is the sum of originated loan amounts in the 2024 HMDA register for the lenders giving that reason, in dollars as HMDA reports them. Methodology

Two sets of banks give different reasons

The same question put to European banks comes back differently. Among the 93 EBA reporting banks that filed an ultimate-parent exception, the most common reason is No known controlling person (51), while the top lenders' most common reason, Parent does not consolidate, is given by 37 of them. The lenders are more often controlled by an entity that does not consolidate; the European banks more often report that nobody is known to control them.

Source: GLEIF, Reporting Exceptions v2.1 format Ultimate-parent exceptions only. The direct-parent question is answered separately, and the two do not have to agree. Methodology

Direct and ultimate are two questions

GLEIF defines the direct accounting consolidating parent as the lowest-level entity that prepares consolidated statements consolidating the child, and the ultimate as the highest. Where a group is one company deep those are the same entity; where it is not, they are not, and the file says how often. Worldwide, 118,364 entities carry both an active direct and an active ultimate parent edge, and 29,538 of them, 25.0%, name a different entity for each. Among the top lenders it is 13 of 70; among the EBA banks 8 of 20. A further 12 entities across the two universes answer one question with a parent and the other with an exception. Merging the two questions, or walking the direct chain upward to reconstruct an ultimate parent rather than reading the one the filer reported, would quietly overwrite all of that.

Source: GLEIF, Relationship Record (RR-CDF) v2.1 format | GLEIF, State Transition and Validation Rules v2.8.5 Every entity appears exactly once per population. The top lenders' direct-parent question is answered by 71 parent edges and 227 exceptions, against 78 and 220 for the ultimate question. Methodology

Where the parents sit

Once an ultimate parent is named it can be placed. Of the 78 named ultimate parents of the top lenders, 4 sit in a different country from the lender itself, 5.1%. Of the 23 named ultimate parents of the EBA banks, 13 do, 56.5%. Both denominators are small, 23 in the European case, so this is a description of these banks and not an estimate of anything wider. Country is the country of the legal address GLEIF holds, which is where the entity is registered rather than where it does business.

Source: GLEIF, LEI golden copy files Up to six jurisdictions per universe, as ISO 3166-1 alpha-2 codes, ordered by entity count then code. The lenders' named ultimate parents sit in 3 jurisdictions in all, the EBA banks' in 11, so a list of six leaves some out where that number is larger. Methodology

Parents that appear more than once

5 ultimate parents named by more than one entity in the same universe. Nothing here is a controlling stake or a merger: it is the consolidation perimeter each filer reported, which is the definition GLEIF uses.

Source: GLEIF, LEI golden copy files Names as GLEIF's Level 1 file records them. 101 of the named ultimate parents in these two universes resolve to a Level 1 record and 0 do not, so no row here is a blank standing in for an unresolved LEI. Methodology

What else the relationship file holds

Consolidation is not the only relationship GLEIF records, and it is not even the most common one. Across the 481,443 active relationship records in this vintage the largest single type is fund managed by, at 148,089 records, which is a fund naming its management company rather than a parent naming a subsidiary. Only the two consolidation types feed the accounting above. 3,039 active records point an entity at itself, and 3,021 of those are of the fund managed by type, which is what a self-managed fund looks like in this file. They are counted and left in the type totals below, and 0 entities in the two platform universes are their own reported parent.

Source: GLEIF, Relationship Record (RR-CDF) v2.1 format Type labels are shortened from GLEIF's enum tokens. Fund management, sub-fund and feeder relationships describe fund structures, not accounting consolidation, and none of them is read as a parent here. Methodology

What is left unaccounted

5 entities across the two universes report no ultimate parent and filed no exception, and they divide cleanly. 3 of them are not LEIs at all: they sit in the identifier column of the EBA table but fail ISO 17442, the standard that fixes an LEI at twenty characters ending in two check digits, and not one of them has a record in GLEIF's file. Every EBA identifier that is a valid LEI is accounted for, 116 of 116. The other 2 are real LEIs belonging to lenders in the curated top-lender table that have simply not answered, and each of them carries "Credit Union" in its registered legal name. GLEIF collects parent data at issuance, renewal or transfer, whichever comes first, and began collecting it in May 2017, so a record that has not been renewed since then can carry neither a parent nor an exception; 1 of these 2 sits on a registration that is no longer issued.

Source: GLEIF, LEI golden copy files These entities are named because they are already named on this platform, on the lending and EU bank pages. The check-digit test that rejects 3 of them is validated against GLEIF's own file first: of 3,384,122 published LEIs, 256 fail it and 0 of those are on a registration GLEIF has not annulled or marked duplicate. Methodology

Sources

• Global Legal Entity Identifier Foundation, LEI golden copy files. The Level 1 entity file, the Level 2 relationship record file and the reporting exception file of one daily publication, 2026-07-25 00:00 UTC. GLEIF publishes LEI data under the CC0 1.0 Universal public domain dedication. This module reads three files and republishes none of them.
• GLEIF, Relationship Record (RR-CDF) v2.1 format. The normative schema, for the relationship-type and relationship-status definitions quoted above.
• GLEIF, Reporting Exceptions v2.1 format. The normative schema, for the exception categories and the verbatim reason-code definitions, including the deprecation of five codes from 1 March 2022.
• GLEIF, State Transition and Validation Rules v2.8.5. The reporting duty itself, the branch exemption, and the definitions of the direct and ultimate accounting consolidating parent.
• LEI Regulatory Oversight Committee, collecting data on direct and ultimate parents, March 2016. The policy the reporting duty and the option to decline with a public reason both come from.
• CFPB and FFIEC, HMDA Loan/Application Register. The 2024 origination volumes used to weight the lender accounting. A US federal government work.