Why Are Banks Disappearing? Too Big to Fail Is Failing the Free Market

In 1984 the United States had 17,901 federally insured banks and thrifts. Today it has 4,238. Most didn't fail. They were absorbed, usually during a panic, usually with official help, and usually by the same few names.

A story of US and Canadian bank consolidation, told through the crises that drove it, from 1895 to 2023. Same pattern every time. And a look at why fewer banks means fewer doors between the state and your savings.

In September 1984, Congress held hearings on the government's rescue of Continental Illinois, then one of the country's biggest banks, with about $40 billion in assets. The FDIC had put in $2 billion and taken an 80 percent stake. Comptroller of the Currency Todd Conover told the committee regulators had no way to let the nation's largest banks fail. Commentators took that to mean the eleven biggest.

Connecticut Congressman Stewart McKinney put it more plainly:

"Let us not bandy words. We have a new kind of bank. It is called too big to fail. TBTF, and it is a wonderful bank."

Here's the part nobody mentioned in that hearing room. 1984 was also the year American commercial banks peaked at 14,496, the high-water mark since World War II. Count savings banks and thrifts too and the FDIC insured 17,901 institutions that year.

By the end of June 2026, the FDIC counted 4,238.

Do the math and it works out to nearly one bank gone every single day. For four decades.

Up north the story is shorter, because there was less to consolidate. Canada's six largest banks hold about 93 percent of the country's banking assets. In October 2025 the Bank of Canada's own Senior Deputy Governor, Carolyn Rogers, called the system an "oligopoly" and added, "To state the obvious, this is a very high level of concentration."

The usual explanation is efficiency, technology and scale. Some of that is real. But line up the moments when the most banks disappeared and a different picture shows up: consolidation happens in bursts, during crises, and the same kind of buyer walks away with the pieces every time.

Past the headline: a pattern, not a trend

The FDIC's own research is blunt about where the banks went. Of the 15,432 institutions that disappeared between 1984 and 2011, only about 17 percent failed. Nearly half merged into unrelated banks, and another third were folded into their own holding companies. Over the same stretch, community banks' share of industry assets fell from 38 percent to 14.

So failure isn't the main way banks die. Absorption is. And absorption speeds up in a panic, because a panic is when good assets get cheap and the rules get flexible.

1895 to 1907: the panics that built the Fed

The pattern starts earlier than most people think. By early 1895, two years into a depression, the US Treasury's gold reserve had drained to about $9 million. President Cleveland turned to J.P. Morgan and August Belmont, the Rothschilds' man in America. On February 8 they signed a contract: their syndicate would deliver about 3.5 million ounces of gold in exchange for roughly $62 million in 30-year Treasury bonds, priced at 104.5. Within a week the public was buying those bonds at 112. Congress and the press were furious.

Economic historians Jon Moen and Mary Rodgers say the deal made Morgan "the de facto lender of last resort for the U.S." A private bank, partnered with Europe's most famous banking family, rescued the government in a panic and got paid well for it. Hold onto that picture.

Twelve years later, Morgan ran the rescue again. On October 22, 1907, the Knickerbocker Trust Company in New York ran out of cash at 12:30 in the afternoon, after paying out $8 million in about three hours. Morgan organized the rescue of the panic. Not of the Knickerbocker. It was left to fail while the firms Morgan judged worth saving got support.

One independent researcher who has spent years tracing this pattern argues Morgan "used it as an excuse to eliminate his banking competition (the Knickerbocker Trust) and rescue his banking associates." He also claims the rumours that spread the run were planted by Morgan partner George W. Perkins. Contested, but not invented: in 1912 testimony, the president of the Trust Company of America said the run on his firm began after the New York Times reported Perkins saying help was coming to it. The Times disputed that. Weigh it how you like.

What isn't disputed is what came next. In November 1910 a small group boarded Senator Nelson Aldrich's private rail car for what staff were told was a duck hunting trip to Jekyll Island, Georgia. They used first names only. Among them: Henry Davison, a partner at J.P. Morgan; Frank Vanderlip, president of National City Bank; and Paul Warburg of Kuhn, Loeb. By the Federal Reserve's own account, the attendees didn't admit the meeting happened until the 1930s. Their plan became the blueprint for the Federal Reserve Act of 1913.

So the man who'd been the private lender of last resort ended up with his partner helping design the public one. Put that way, it's hard to call the Fed a referee.

Meanwhile, in Canada

Canada was running a different experiment. It had no central bank at all. Its chartered banks issued their own notes and branched coast to coast, and by most accounts it worked: no US-style panics, and no bank failures during the Great Depression.

It consolidated anyway. Chartered banks peaked at 41 in 1886. The Bank Act of 1890 raised the capital needed to open one, and new entry nearly stopped. The Bank Act of 1900 made mergers easier, and 27 followed by 1926. By 1935 Canada had ten banks.

That same year the Bank of Canada opened. Not because the private system had collapsed (it hadn't), but, as Cato's banking historians put it, out of "some combination of nationalism and wishful thinking about what a central bank could do to end the Great Depression."

1984: too big to fail becomes policy

Once regulators said the biggest banks wouldn't be allowed to fail, they handed them something worth more than any branch network: cheaper deposits. If you have more than the insured limit, why keep it at a small bank that can fail when a big one effectively can't?

That's a subsidy, and subsidies shape markets. The years that followed brought the worst wave of failures since the Depression, as the savings and loan crisis and the early-90s bust took out 2,555 institutions, mostly small ones. Then came the legal plumbing for bigger banks. The Riegle-Neal Act of 1994 allowed full interstate branching, with one guardrail: no bank could grow past 10 percent of national deposits through acquisitions. Remember that number.

Canada ran its own version. In 1967 Ottawa created federal deposit insurance, and not for safety: no Canadian bank had failed since 1923. The goal was to help trust and loan companies compete with the big banks. Instead, 43 insured institutions failed, including Canada's first two bank failures in 62 years, in 1985. By 1990 the big banks faced fewer retail competitors than before.

Then, in December 1998, Finance Minister Paul Martin blocked two mega-mergers (Royal Bank with Bank of Montreal, CIBC with TD), citing too-big-to-fail risk among his main reasons. Canada's banks came through 2008 in better shape than most, and that refusal often gets the credit.

Which makes the next part a little awkward.

2008: the crisis that cleared the field

In March 2008 JPMorgan bought Bear Stearns, with the Federal Reserve financing $29 billion of Bear's riskiest assets to make it happen. On September 25, regulators seized Washington Mutual, the biggest bank failure in US history at $307 billion in assets, and sold its banking operations to JPMorgan that night for $1.9 billion. Bank of America took Merrill Lynch and Countrywide. Wells Fargo took Wachovia.

Notice the name that keeps coming up. In 1895 and 1907 the House of Morgan ran the rescues. A century later, the bank that carries its name was the one buying.

Canada's official line is that its banks didn't need bailouts. In 2012 the Canadian Centre for Policy Alternatives went through the records and found that support from the Bank of Canada, CMHC and the US Federal Reserve peaked at $114 billion in March 2009. At points, support for CIBC, BMO and Scotiabank exceeded those banks' entire market value. For CIBC it reached 148 percent. The government called it "liquidity support." The report's author, economist David Macdonald, said, "it looks like a bailout to me."

In 2013 Canada's bank regulator, OSFI, formally designated the Big Six as domestic systemically important banks. That's just a fancier way of saying too big to fail.

Then something quieter happened on both sides of the border: new banks stopped being born. From 1960 to 2006 the US averaged about 185 new bank charters a year. From 2010 to 2024, fewer than seven. A new community bank now typically needs $20 to $25 million just to open, which is Canada's 1890 Bank Act with a bigger number.

2023: the cap that wasn't

In March 2023 Silicon Valley Bank failed in about two days, followed by Signature Bank. Deposits ran toward the biggest institutions. The same researcher noted that while prominent venture capitalists told startups to pull their money from SVB, "mega institutions such as JPMorgan Chase sought to convince some SVB customers" to move over. He goes as far as saying "the fall of SVB was engineered . . . by the Fed." Design? Or the predictable result of the fastest rate hikes in forty years gutting SVB's bond portfolio? Either way, the deposits ended up in the same place.

On May 1, 2023, regulators seized First Republic Bank and sold it to JPMorgan. Remember that 10 percent cap? JPMorgan already held about 13 percent of US deposits. The cap has an exception for buying failed banks, so it didn't apply. The FDIC agreed to share the losses, at an estimated $13 billion cost to the deposit insurance fund.

So the biggest bank in the country got bigger, past the legal limit, with a $13 billion assist. Because it was an emergency. (We've seen the same reflex in private credit, covered in Wall Street Spent a Decade Marking Its Own Homework. Privatize the gains, socialize the panic.)

Today the four largest US banks hold 38.9 percent of all domestic deposits. JPMorgan alone holds $2.8 trillion.

Now: no crisis required

Here's why it feels like it's speeding up. It is.

In May 2025 the OCC rescinded its 2024 merger policy and brought back expedited reviews. Approvals that could drag on for 18 months to two years now often come through in three or four, according to bank M&A advisers. US banks announced 181 deals in 2025, the most since 2021. Fifth Third closed its purchase of Comerica in February 2026, creating the country's ninth-largest bank.

Canada's pace looks small, but its starting point was six. RBC bought HSBC Bank Canada for $13.5 billion in March 2024. National Bank bought Canadian Western Bank in February 2025. And on November 1, 2026, Laurentian Bank is scheduled to be split between Fairstone Bank and National Bank, ending a 180-year run as a listed company.

For more than a century, consolidation needed a crisis for cover. It doesn't anymore.

Why this was always the destination

Jesús Huerta de Soto, whose Money, Bank Credit, and Economic Cycles is about as thorough an Austrian treatment of banking as exists, explains why this keeps happening. Banks that lend out money depositors think they can withdraw at any moment are always one rumour away from a run. In his words:

"When the principles which should govern the irregular-deposit contract are violated, such acute and inescapable effects appear that private bankers soon realized they needed to turn to the government for an institution to act on their behalf as lender of last resort and provide support during stages of crisis, which experience demonstrated to be a recurrent phenomenon."

(I had to read that sentence twice too. It's worth it.)

The central bank wasn't forced on bankers against their will. It was something bankers needed, and asked for (some of them on a 'duck hunting trip'). Morgan had already shown in 1895 what a lender of last resort was worth to whoever played the part. De Soto calls the ability to operate with fractional reserves a privilege "granted in the past by governments for reasons of mutual interest," and an "attack by government authorities" on depositors' property rights.

Once you see it that way, consolidation stops looking like a market outcome. A system that needs a lender of last resort will rescue whoever is most dangerous to let fail, and that mostly comes down to being big. So size gets subsidized, rescues go to the biggest buyers, and the cost of entry keeps rising until no one new shows up. De Soto goes further, treating central banking as the same calculation problem that sank socialism: one agency deciding how much money and credit an entire economy gets. Planned systems end up with a few big, well-connected players.

Fewer banks, fewer doors

Why should you care how many banks there are? Because every bank is a door between the state and your money, and fewer doors are easier to guard.

In February 2022 the Canadian government invoked the Emergencies Act, and within days banks had frozen 257 accounts holding about $7.8 million. No judge signed off. Police sent lists to the banks, and the banks complied. In a country where six institutions hold 93 percent of the assets, that took a handful of phone calls. Two federal courts have since ruled the freezes unconstitutional. We wrote about why financial privacy matters before you need it in Jump out of the pot.

The US version is quieter but real. Operation Choke Point pressured banks to drop legal but disfavoured businesses in the 2010s, and in August 2025 the White House issued an executive order on "debanking," conceding the practice is real enough to need one.

Now look ahead. The same researcher warns that banking crises create political cover for central bank digital currencies, and that if nothing checks them, they "will reach global ubiquity," running on "real-time surveillance and algorithmic permission of all transactions." Maybe, maybe not. But which system makes programmable, permissioned money easier to roll out: 14,496 independent banks, or four that already hold almost 40 percent of the deposits?

The one system that can't be merged

So what's the alternative? A smaller set of better-behaved giants?

Bitcoin answers de Soto's problem at the root. There's no fractional reserve in a wallet you control. Your coins are there or they aren't, and anyone can verify the total supply. There's no lender of last resort because there's no last resort to need. No panic can hand your holdings to a bigger institution overnight, because there's no institution in the middle.

Here's the part I keep coming back to. In the system we just walked through, the number of banks has fallen for forty years. In Bitcoin, every person who takes their coins into self-custody becomes, in effect, a 100 percent reserve bank of one. That count goes up every day.

It doesn't happen automatically, though. Bitcoin left on an exchange is just another IOU, and those consolidate exactly like banks do. The protection only works if you actually hold the keys.

Not your keys, not your coins. Not your bank, not your rules.

Every coin you buy at Bitcoin Well goes straight to your own wallet, so you're the one holding it.

A note on sourcing

Bank counts and the FDIC's own numbers. The 1921 peak (30,456), 1984 post-war peak (14,496) and 2020 figures come from the St. Louis Fed's "Slow, Steady Decline in the Number of U.S. Banks Continues" (2021). The Q2 2026 count (4,238; 36 mergers, 4 new charters, 1 failure) is from the FDIC Quarterly Banking Profile via the ABA Banking Journal. The 1984–2011 exit breakdown and community bank asset share are from the FDIC's 2012 Community Banking Study, chapter 2. Deposit shares are from Q2 2026 call report data compiled by Banksparency.

The crises. The 1895 gold bond deal: Jon R. Moen and Mary Tone Rodgers, "J. P. Morgan: The Making of a Private Lender of Last Resort, 1882 to 1896" (Economic History Society paper), with the February 8, 1895 contract date and $62 million figure from the Dictionary of American History's "Morgan-Belmont Agreement" entry and American Heritage's "The Golden Touch." Continental Illinois and McKinney's quote: Federal Reserve History. Knickerbocker's suspension: finhist.com's archive of 1907 press accounts. The Thorne testimony on the Trust Company of America run: the company's history, citing 1912 testimony. Jekyll Island: Federal Reserve History's essay on the conference. WaMu: CNN Money, September 25, 2008. First Republic and the 10 percent cap: BauerFinancial's "When a Cap Is Not a Cap."

The researcher. The 1907/2023 parallels and CBDC warnings come from independent journalist James Corbett, "Party Like It's 1907" (April 2023) and New World Next Week episode 512 (March 17, 2023).

Canada. Bank counts 1886–1935 and the Bank Acts of 1890 and 1900: Cato Institute, "Entry in Canadian Banking, 1870–1935." The Bank of Canada's founding and Depression-era record: Cato, "What You Should Know About Free Banking History." CDIC and the post-1967 failures: Cato, "Unnecessary Evil: How Canada Ended Up Insuring Bank Deposits." The $114 billion figure: Canadian Centre for Policy Alternatives, The Big Banks' Big Secret (David Macdonald, April 2012). D-SIB designation: CBC, 2013. Rogers' "oligopoly" remarks: The Logic, October 9, 2025. Laurentian, HSBC Canada and CWB deal details: company releases and Wealth Professional.

The merger surge. OCC news release NR 2025-44 (May 8, 2025); Banking Dive's 2026 M&A outlook; Fifth Third investor release (February 2, 2026); American Banker and PYMNTS on de novo charters (September 2026).

The Austrian anchor. Jesús Huerta de Soto, Money, Bank Credit, and Economic Cycles (Ludwig von Mises Institute, English edition). The lender-of-last-resort and "mutual interest" passages are from the Introduction (pp. xxv–xxvi); the "attack by government authorities" line is from p. 810.