Chicago Makes the Banks That Hold Its Money Report Their Lending. Four Files, 337,429 Rows — and Five Different Spellings of BMO.

Four files, one ordinance

The four files as retrieved on September 16, 2026. Every one was last updated December 15, 2025.

File	Rows	Bank names as filed	What a row carries
Commercial and consumer lending	30,571	41	A year, a bank, a zip or census tract, a loan count and a combined amount
Residential lending	100,937	42	A year, a bank, a census tract, an outcome code, a purpose, an amount and a date
Savings and checking accounts	203,922	41	A year, a bank, an area, an account count and a combined balance
Depository locations	1,999	40	A year, a bank and an office record, counted here without reading its location fields

The reporting is not voluntary: the files say each is published because the Municipal Code requires certain banks to report and the Comptroller to make the reports public. What follows is what those reports contain, read across every row, and what stands between them and a simple total.

Forty-seven bank names, grouped here into twenty

Every institution the four files spell more than one way. The grouping is ours; the names are the files’.

Institution	Spellings in the files	The names as filed
BMO	5	BMO; BMO Bank N.A; BMO HARRIS BANK, N.A.; BMO Harris Bank National Association; BMO Harris, N.A.
Fifth Third	5	Fifth Third; Fifth Third Bank; Fifth Third Bank N.A; Fifth Third Bank National Association; Fifth Third Bank, N.A.
Citi	4	Citi; Citibank; Citibank N.A.; Citibank, N.A.
JPMorgan	4	JPMorgan; JPMorgan Chase & Co.; JPMorgan Chase Bank, N. A.; JPMorgan Chase Bank, N.A.
Bank of America	4	Bank of America; Bank of America N.A.; Bank of America, N.A.; Bank of America, National Association
Liberty Bank	3	Liberty Bank; Liberty Bank and Trust; Liberty Bank and Trust Co.
PNC	3	PNC Bank; PNC Bank, N.A.; PNC Bank, National Association
Wintrust	3	Wintrust; Wintrust Bank; Wintrust Bank, N.A.
Albany Bank	2	Albany Bank & Trust Co NA; Albany Bank and Trust Company, N.A.
Huntington	2	Huntington; Huntington National Bank
Old National	2	Old National Bank; Old National bank
Wells Fargo	2	Wells Fargo; Wells Fargo Bank, National Association

Across the four files, 47 distinct bank names appear. Folding the obvious variants leaves 20 institutions, and 12 of those are written more than one way: BMO and Fifth Third each appear under five names, and Citi, JPMorgan and Bank of America under four apiece. Each of the four files folds to 18 institutions, and the four together to 20, even though the names they file run from 40 to 42, so the inconsistency is easy to miss and hard to work around: a reader adding up one bank’s lending has to decide, unaided by the file, whether “BMO,” “BMO Harris, N.A.” and “BMO Harris Bank National Association” are one reporter. This page groups them, and says so: a name can belong to a parent company or to a bank subsidiary, and the file does not distinguish them.

One outcome, five forms, and 1,491 blanks

All 100,937 residential records by what the outcome column holds. The codes are the file’s; the grouping of spellings is ours.

Outcome as filed	Records	How the file writes it
Code 1, the file’s label: loan originated	61,174	1 (46,718); 1-Loan originated (14,456)
Code 3, the file’s label: application denied	19,419	3 (17,849); 3-Application denied (1,570)
Code 6, the file’s label: purchased loan	17,669	6 (11,450); 6-Purchased loan (6,219)
Codes 2, 4 and 5, which the file never labels	1,091	2 (119); 4 (880); 5 (92)
Words with no code	93	Loan Originated (85); Application Denied (3); Approved/Not Accepted (2); Orginated (2); Originated (1)
Blank	1,491	no value at all

The residential file records an outcome on all but 1,491 of its records, and not in one form. Code 1 appears 46,718 times as a bare “1” and 14,456 times as “1-Loan originated”; denials split 17,849 to 1,570 the same way. Ninety-three records skip the code and write words instead — “Loan Originated” on 85, “Application Denied” on three, “Approved/Not Accepted” on two, “Originated” on one, and “Orginated,” misspelled, on two. Another 1,491 records leave the column empty, and 1,091 carry codes 2, 4 or 5 that the file never spells out anywhere. Of the 80,593 records carrying outcome code 1 or code 3, 24.1 percent are denials — a figure about records, not about people, since the file counts filings and a purpose column that is blank on 20,038 of them.

Most of the loans are consumer loans; most of the money is not

The commercial and consumer file’s own three labels, which are two kinds of lending written three ways.

Label in the file	Rows	Loans	Amount	Share of dollars
Commercial Lending, capital L	9,807	38,389	$79,027,442,341	69.6%
Commercial lending, small l	4,122	18,594	$27,836,879,997	24.5%
Consumer Lending	16,642	348,779	$6,724,284,616	5.9%

The commercial and consumer file keeps both kinds of lending in one place, under a label capitalized two different ways. Folded, commercial lending is 13,929 rows, 56,983 loans and $106.86 billion — 14.0 percent of the loans and 94.1 percent of the dollars. Consumer lending is the mirror image: 348,779 loans, 86.0 percent of the count, carrying $6.72 billion, 5.9 percent of the money. Any reading of this file that does not separate the two is measuring two different things at once.

What the files do let you count

Each column computed from one file’s own rows for that year. The account count leaves out rows that carry no count — JPMorgan’s aggregate filings in 2023 and 2024, described below. Row and loan counts move for reasons the files do not explain, so the columns are read down, not across.

Reporting year	Commercial and consumer loans	Commercial and consumer amount	Residential records	Residential amount	Accounts, where a count is filed	Account balances	Office records
2021	77,327	$16,821,082,298	26,248	$8,212,742,741	1,710,200	$49,467,204,283	289
2022	113,787	$33,229,710,734	29,558	$10,591,510,790	2,326,561	$84,044,884,598	468
2023	120,149	$17,863,275,679	22,324	$7,665,177,971	2,238,932	$70,914,582,764	416
2024	14,788	$25,124,083,651	12,674	$4,619,239,563	2,376,039	$69,600,708,630	493
2025	79,711	$20,550,454,592	10,133	$4,348,594,776	1,746,086	$48,922,843,767	333

Read one column at a time, the series are legible. Commercial and consumer lending reports $113.59 billion across 405,762 loans over the five years, with the loan count falling from 120,149 in 2023 to 14,788 in 2024 while the dollars rose. Residential records fall every year after 2022, from 29,558 to 10,133. Account balances peak at $84.04 billion in 2022 and land at $48.92 billion in 2025. Office records run between 289 and 493 a year. Each of those movements could be a change in lending or a change in reporting, and nothing in the files says which.

Why one year holds nearly eight times the account rows

The accounts file jumps from 18,487 rows in 2023 to 142,640 in 2024 — 7.7 times as many — and back to 14,158 in 2025. One reporter accounts for it: Citibank filed 126,337 of the 2024 rows, 88.6 percent of that year, where the next-largest filer lodged 2,660. The labels move too — 2023 has 50 rows reading “Saving and Checking Account (Aggregate)” and 2024 has 82 reading “Savings and Checking Account” — and the zip column is empty for every row in 2022 and 2025, while 2023 carries 50 distinct zips and 2024 carries 48. A year-to-year comparison of row counts in this file is not on its own a measure of banking activity. The file also carries rows labelled as a savings-and-checking aggregate — 50 of them in 2023 worth $77,150,781 and 82 in 2024 worth $4,971,935,000 — and both sets are JPMorgan’s only filing for those years, so those aggregate rows duplicate none of the itemized filings. Those rows carry no account count, which is why the account-count column above leaves that bank out; its balances sit in the balance column.

What the files confirm, and what they cannot tell you

Confirmed, as reported: for each year and each bank name, the loan counts and amounts, the account counts and balances, the residential outcomes and purposes as coded, and the number of office records. The columns read here do not show: any borrower, customer or account holder, none of whom the files identify; where an office is, since the address column was never selected; what any bank charges, since no interest rate is published against a name here; whether a bank met any city requirement, which these files do not score; or why any figure moved. Nor can the four files be joined on a bank name without the folding this page discloses.

Where the record lives

The four files sit on the city’s data portal: commercial and consumer lending, residential lending, savings and checking accounts and depository locations. Each was last updated December 15, 2025, and every figure here was computed from them as retrieved on September 16, 2026.

The bottom line

The ordinance has produced a record: four files, 337,429 rows, five reporting years, and hundreds of billions of dollars described. What it has not produced is a record that adds up on its own — 47 bank names for 20 institutions, an origination filed in five forms, 1,491 records with no outcome at all, and a year in which one bank’s filing practice sets the shape of the file. Both things are true at once. It is the same pattern this desk found in the city’s MBE and WBE payment record, which is complete in form and stops reporting in March 2024.

Two more questions the files can answer

How much of the residential file is a purchase rather than a decision? 17,669 records carry the file’s purchased-loan code, 17.5 percent of all residential records. And how many institutions do those names fold into? Eighteen in each file and 20 across all four, by the folding this page discloses, from 47 names as filed.

• Sources: the City Comptroller’s four Lending Equity files: commercial and consumer lending (4t5u-cj3n, 30,571 rows), residential lending (b77m-uuhb, 100,937), savings and checking accounts (2tsv-5s43, 203,922) and depository locations (g5wi-brhg, 1,999). All four were last updated December 15, 2025 and retrieved September 16, 2026.
• How they were counted: by group-by queries run over every row of each file, with each file’s totals checked back against its own row count, so no row is dropped silently.
• Our folding of names: bank names were upper-cased, stripped of punctuation and of suffixes such as N.A. and National Association, then matched to a short list of institution stems. Both counts are published: 47 names as filed, 20 institutions after folding.
• The grain was checked: the accounts file carries rows labelled as a savings-and-checking aggregate; a year total would double-count if the banks filing them also filed itemized checking and savings rows. Grouping every row of that file by year, bank and description shows they do not: those rows are one bank’s only filing in each of the two years they appear, and they carry no account count, which is why the account totals leave that bank out.
• Rows are records: a residential row is one filing with one outcome value — coded, written out or blank — not one person; an accounts row is a bank’s report for an area, not a customer. Nothing here counts people.
• Not here: no borrower or customer detail, no office address or coordinate, no interest rate against a bank name, no compliance judgment, and no comparison of row counts across years where the files change shape.

Computed by KCM Desk from the City Comptroller’s four Lending Equity files, retrieved September 16, 2026. Human-guided and edited — about this desk. If you spot an error, corrections come first.

Related Keep Chicagoland Moving coverage

• How Chicago Works, by the Numbers
• Chicago’s Record of Payments to Minority- and Women-Owned Firms on Procurement-Department Contracts Stops in March 2024 — and Its Last Period, at 24.7 Percent, Is the Lowest of the 18 in the File
• Chicago’s Contractor Diversity-Program File Holds 400 Reports From 266 Filers — None at All for Calendar 2024, and Eight for 2025
• Chicago Paid $5.72 Billion on Contracts in 2025. Ten Payee Names Took 30 Cents of Every Dollar — Barely Changed From 2019.