Your bank is far likelier to be sold than to fail, and the sale is the event that moves your money

A bookkeeper wrote to me in June with a 2 line question, and I answered a different question than the one she had asked, which is something I do more often than I would like and nearly always when the question sounds like one I have answered before. Her bank had been bought. I sent her 4 paragraphs about deposit insurance, the 250,000 dollar limit and ownership categories, which is the right material when a bank fails. Her bank was not failing at all. I had assumed for years that a sale and a failure sit close enough together to share one answer, and I gave her bad advice in a confident voice.

What annoys me is how ordinary the letter she forwarded was. I have read a lot of them now. They open with something like “an exciting new chapter for our customers” and they close with a phone number that nobody answers on a Friday. The word merger usually appears twice. The words routing number appear nowhere in it. Nothing in the letter tells an owner what to do on the Monday, and nothing in it is untrue either, which is the part that makes the whole genre so difficult to argue with and so easy to file in a drawer.

So I went looking for the ratio between the 2 events, because I wanted to know how badly my emphasis had been off. The FDIC publishes both lists. Neither list publishes a ratio, and the counting took most of an afternoon.

What happens to my bank account when banks merge is 2 questions rather than 1. The insurance question has a federal answer. The operations question has a private answer, it carries a deadline, and nobody sends a reminder about it.

The number that changed my emphasis

On 24 August 2026 the register held 4,245 active insured institutions. This paper ran the same query on 28 July and got 4,255. That is 10 banks gone in 27 days, and not one of the 10 failed.

117 charters ended in 2026 up to today. The failure list for the same 8 months holds 2 names, and I find that gap hard to read past. Community Bank and Trust West Georgia in LaGrange closed on 1 May with 305.7 million dollars of assets. Metropolitan Capital Bank and Trust in Chicago closed on 30 January with 261.2 million. Everything else inside that 117 was a sale, a merger into a sister charter or a quiet wind down, and none of it put a receiver in a lobby.

The ratio this year is 58 to 1. I pulled the earlier years expecting 2026 to be the odd one, and it is not odd at all: 2025 recorded 152 exits against 2 failures, 2024 recorded 106 against 2, 2022 had 142 exits with no failures and 2021 had 172 exits with no failures either, and even 2023, the year everybody remembers, with 209 billion dollars at Silicon Valley Bank in March, 110.4 billion at Signature 2 days later and 212.6 billion at First Republic on 1 May, came to 5 failures against 130 exits.

I want to be careful about what a ratio like that means and what it does not. It does not make a failure impossible at your bank. It means the odds are lopsided in a way our advice has never reflected, and the shape has held in every year I can check, through a quiet stretch and through the loudest banking spring in a generation. The failures are the ones that come with photographs attached. The sales arrive as a letter in an envelope with a new logo on it, and they are the ones that actually reach into the plumbing of a company. I have written the insurance piece more times than I have written this one, and that was a comfortable choice rather than an honest one.

The names leaving are not the small ones

Comerica Bank held 80.05 billion dollars and its charter ended on 1 February, Synovus Bank held 61.2 billion and ended on 2 January, Cadence Bank held 53.5 billion and ended on the same day as Comerica, and FirstBank ended on 18 June holding 29.4 billion, which is 4 institutions that between them looked after more than 224 billion dollars of somebody else’s working capital. None of the 4 was in trouble. Size is not the protection people assume it is.

Under those sit First Foundation Bank at 11.3 billion, Stellar Bank at 10.9 billion, Flushing Bank at 8.9 billion and MidWestOne Bank at 6.3 billion. A company could have called any of them its main bank in January. By August none of them exists as a charter.

I cannot tell you from that list when a single customer actually moved. The register records the day a charter stopped existing. Accounts convert on a different day, and the gap runs from one weekend to the better part of a year. I have not found a published median for that gap, and I went through the FDIC site and the Federal Reserve services site looking for one.

Six months of cover, and a cliff at the end of it

The rule that governs the money is 3 lines long and sits at 12 CFR 330.4. It says that “the separate insurance of deposits assumed continues for six months from the date the assumption takes effect or, in the case of a time deposit, the earliest maturity date after the six-month period”.

I would read that with a calendar open. Take a business holding 200,000 dollars at bank A and 200,000 at bank B. The company is fully insured today. It is still fully insured 5 months after the deal closes, because the 2 balances are treated as though they sat in 2 separate institutions. On the first day of month 7 they are a single balance of 400,000 dollars at a single insured bank, the standard limit of 250,000 applies to the lot, and 150,000 dollars is uninsured.

Nothing arrives in the post to mark that date. I keep thinking about how quiet the mechanism is. The protection is generous, it is automatic, and it expires on a day that appears in no letter and no statement. I find that stranger than any number on this page.

Certificates run on a separate clock. A 18 month certificate bought before the deal keeps its own coverage until the first maturity date that falls after the 6 months expire, which can push the cliff out by a year. The FDIC states the general limit as “250,000 dollars per depositor, per insured bank, for each account ownership category”, and the merger clause is a temporary exception to it rather than a permanent gift.

The parts that move on conversion weekend

Insurance at least has a published rule. The operational side has none, and that is where a company loses a week, because every system that has ever been told your account details keeps a copy of them, and those copies live at your vendors, at the payroll bureau, in the tax portals and inside software that somebody configured 3 years ago and left. The routing number is the first question. It has 2 possible answers, and some acquirers retire the old number at conversion while others carry it for a while. One bank told a client of mine, in writing, that “the routing number will be retired at systems conversion”, which is the sentence you want on paper.

The account number is the second question. A change there breaks every debit a vendor set up against the old one. I would pull 90 days of statements and write down every originator that takes money out, because that list is always longer than memory says it is. Payroll, the insurance carrier, the merchant processor holding your funding account, EFTPS, the state tax portal, the lease on the card terminals and the 2 subscriptions nobody remembers signing.

It sounds petty until a payroll file bounces on a Thursday. Somebody has to ring every vendor by hand after that. The week is unpleasant and it is entirely avoidable. I have watched careful people lose it anyway.

Online banking is the third question. Users, entitlements and dual control rules do not survive a platform change. The person who leaves the company during a conversion is the person whose access nobody removes, and I find that pattern uncomfortable every time I see it. Positive pay rules and their cutoff times are rebuilt from scratch on the new system, and a cutoff that shifts by 2 hours turns a same day wire into a next day wire.

The fee schedule and the deposit agreement come last in the letter and first in my order. The deposit agreement decides your liability for an unauthorised transfer, because Regulation E stops at personal, family or household and a company account sits outside it. A new owner means a new agreement. The notice window inside it is often 14 days rather than the 60 you remember from your personal account.

The order I would work in

In the week of the announcement I would ask for 3 things in writing and keep the answers: the planned conversion date, a straight answer on whether the routing number survives it, and the place where the new deposit agreement and fee schedule can be read today. A bank that cannot answer the first 2 in August will answer them in September. Asking early also puts you in the file as a customer who is watching.

About 30 days before conversion, build the debit list from statements, put a name against every line and send the new details in one batch rather than in 30 phone calls. Move the smallest vendor first as a test. If the company holds more than 250,000 dollars across the 2 merging banks, this is the month to decide where the surplus goes, and month 7 is the deadline you are working back from.

On the conversion weekend, send yourself 1 dollar by wire and 1 by ACH on the Monday morning, and check the balance and the statement format before anything real depends on either. Export the last 24 months of statements from the old platform in the same week. Archives are the thing that quietly disappears. No bank has ever restored one quickly for a customer with 3 employees.

In the first 30 days after conversion I would compare the new fee schedule against the old one, line by line. The number worth watching is the earnings credit or the interest on the operating balance, which changes most often and is advertised least. Anything that got worse is negotiable in month 1 and settled forever by month 6.

A short digression about routing numbers

A short digression, and it is only half useful. The Federal Reserve keeps a public routing directory, and typing a routing number into it shows which institution owns that number today, which is a different question from the one your statement answers and a more current one, because the directory is maintained so that payments arrive rather than so that a bank looks tidy in public. The check takes 20 seconds. I have used it twice this year to work out that a client was banking with a company that no longer existed under that name. Back to the point at hand.

What I could not establish

The median gap between the day a charter ends and the day customers convert. It is not published by the FDIC, it is not in the Federal Reserve material, and every bank announces its own date in its own letter, which means the only way to build the figure would be to collect announcements one at a time across a year, and I have not done that. My guess is that it clusters between 4 and 9 months for deals of this size. I would not defend the guess. Somebody with access to the conversion calendars could settle it in an afternoon.

Whether pricing gets worse for small business customers after an acquisition. I suspect it drifts rather than jumps, because a jump gives a customer a reason to leave and a drift gives them a reason to complain quietly and stay, and the quiet complaint is worth more to an acquirer than the loud exit costs it. I have not found a study that measures it. It is a guess and I am not going to dress it up.

Whether any of this feels different at a small bank. My instinct is that a community bank conversion is gentler in tone and identical in mechanics, and an instinct is not a finding.

The detail I still cannot get past is the 58 to 1. Everything I have written about business deposits, including 2 pieces in this paper, prepares an owner for the rare event and says almost nothing about the ordinary one, and the reason is not that the ordinary one is harder to write about, it is that insurance comes with a rule and a number while a merger comes with a project plan that belongs to somebody else. That is an embarrassing thing to notice in August. I would rather be embarrassed now than write the insurance piece again in September.