The Rating the Order Forgot
On October 29, the Federal Reserve Board unanimously approved the Bank of Nova Scotia’s purchase of Maple Financial Holdings and its subsidiary, MapleMark Bank of Dallas and Tulsa. The buyer holds about $1.1 trillion in assets, and the target holds about $1 billion. The order finds every statutory factor “consistent with approval.”
The community lending section rests on a single document. It is a July 2024 FDIC evaluation that rated MapleMark “Satisfactory” under the Community Reinvestment Act.
What the order leaves out is the rating before it. The exam completed in November 2022 rated the bank “Needs to Improve,” according to the FDIC’s own 2024 evaluation, which MapleMark posts in its public CRA file. The order gives that earlier exam’s date in a footnote and never mentions its result.
That is a move from the second-lowest CRA rating to a passing grade in about twenty months. S&P Global reported in 2023 that MapleMark and Oakwood Bank both landed “needs to improve” ratings less than a year after withdrawing their merger application.
The Board is not obligated to dwell on old grades. Its stated practice is to weigh an institution’s most recent evaluation. The question is what that evaluation measured.
One Year, Eighty-Eight Loans, No Mortgages
The 2024 exam reviewed small business loans from calendar year 2023 and nothing else. That was 88 loans, 19 of them in Tulsa.
It did not evaluate mortgage lending. Examiners noted the bank only began reporting Home Mortgage Disclosure Act data for 2023 and had no complete year of comparison data.
Mortgage lending is where the lone commenter aimed. The commenter alleged that MapleMark made fewer loans to African American borrowers than to white borrowers in 2025.
The Board’s answer tracks Scotiabank’s. HMDA data are incomplete, MapleMark mostly does commercial lending, and credit scores are not public. The order adds that confidential supervisory information supports approval.
It offers no numbers. No applicant count, no denial rate, no origination total.
Those figures are public elsewhere. According to FFIEC data compiled by BankRegReports, MapleMark took 24 mortgage applications in 2025 and originated 12, about $15.6 million, across four states.
Twelve loans cannot prove a disparity. They also mean the public record holds no regulator’s assessment of this bank’s mortgage lending at all.
The order says the Board reviewed MapleMark’s most recent consumer compliance examination, fair lending included. It does not say when that exam took place or what years it covered.
Satisfactory, With Footnotes
Tulsa is the commenter’s other complaint, and the FDIC’s own tables bear it out. Of 19 small business loans in the Tulsa area in 2023, none went to low-income tracts and three went to moderate-income tracts. That is 15.8 percent against a 24.3 percent benchmark.
Examiners called the geographic distribution “poor.” The order says only that Tulsa lending was reasonable.
A fairness note is due. Low-income tracts hold 4.1 percent of Tulsa area businesses, so 19 loans implies about one such loan, and zero is a small-numbers problem. The moderate-income gap is the sturdier point, and those three loans totaled $272,000.
Scotiabank’s rebuttal is 15 small business loans in the Tulsa area through June 30, 2026, about 13 percent in low-income tracts and 53 percent in moderate-income tracts. That is roughly two loans and eight loans. The order gives no dollar amounts, no start date for the count, and no independent check on the claim that this beats peers.
One inference, resting on the order’s wording: if the count begins at the July 2024 exam, it covers two years and 15 loans, against 19 in 2023 alone.
Community development tells a similar story. The same 2024 exam shows community development lending falling from 4.8 percent of average assets to 2.1 percent, and qualified investments from 1.2 percent to 0.8 percent. Beyond $73,000 in donations, every qualified investment it counted was bought before the review period began.
Scotiabank cited community development loans and qualified investments as proof of the bank’s commitment. The order repeats that without the trend.
What the Buyer Says It Wants
Announcing the deal in May, Scotiabank executive Travis Machen said MapleMark would let the bank offer FDIC deposit insurance to clients. He called that important to its Mortgage Capital Markets business and its deposit growth strategy.
That business is mortgage warehouse lending, short-term funding for independent mortgage banks. Inside Mortgage Finance reported the insurance could help fund it, and HousingWire noted that Scotiabank hired seven JPMorgan executives in Texas in 2024 to build it.
The commenter argued MapleMark would become a funding vehicle for that business. Scotiabank told the Board the commenter described its reasons inaccurately. The order does not say what the reasons are.
Two readings follow. Both are inferences resting on Scotiabank’s announcement and the trade press, not on anything in the order.
If warehouse loans land on MapleMark’s books, “no material changes” is hard to square. If they land elsewhere and MapleMark supplies insured deposits, the commenter’s description fits.
Neither reading is in the record. What the call report data shows is a business bank rather than a neighborhood one: roughly 45 percent of deposits uninsured and about 90 percent held at its Dallas office.
It is also a bank whose June 2026 return on assets was 0.17 percent, in the bottom five percent of its peers, with nonperforming loans up from 0.04 percent to 0.66 percent in a year. Scotiabank called it well run. The Board found its earnings and asset quality consistent with approval.
Inquiries are being directed to the Federal Reserve Board and to Scotiabank. Three questions top the list:
- Where will Scotiabank book its warehouse lending, and what balance sheet growth does it expect at MapleMark?
- When does the 15-loan Tulsa count begin, and what were the dollar amounts?
- When did the consumer compliance exam the Board relied on take place, and did it cover MapleMark’s 2025 mortgage lending?
A missing rating, a one-year record, and a business plan the order never describes make a thin file for a unanimous vote. The Board has spoken. The numbers, so far, have not.
