ASTONISHING piece to see in regular financial press. Maybe the tide is turning and the world is healing?
The basic/most fundamental rationale for a central bank isn't financing the gov deficit (which they don't...ish) or set centralized interest rates, but to support commercial banks when faced with bank runs (=stand ready to lend against their assets, even if distressed, with newly created base money, so stave off the run.) The idea, from Francis Barings to Diamond&Dybvig's Nobel Prize winning article, is that depositors are fickle and might/will run on perfectly healthy banks and that such runs cause real damage to the economy (and might spread through the banking system like wildfire and we all collapse into Great Depression 2.0).
That's everything that's wrong with monetary economics and politics around central banks in one sentence...
The authors of the paper Nangle discusses here used the Chronicling America project, scanned newspapers, then...
they used LLMs to sift these articles further to find pieces about actual bank runs rather than, say, stuff about boats running into river banks, or the local bank manager running for mayor, and to match each report to individual banks.
According to Verner, the research reveals that the idea that runs kill solvent banks is dramatically overstated. It turns out that depositors are actually pretty good at working out how dodgy a bank is, generally making runs on banks that are insolvent or very close to insolvent.
This is incredible. Specifically, — and I'm sparing you a deep dive into debates in financial history and monetary economics here — a big rationale for why central banks must exist is to prevent bad types of bank runs.
Running a bank means it has to shed assets, assets that aren't "finished," or turned into workable projects and thus liqudity, yet. (That's sort of the purpuse of a bank: to liquify illiquid assets, hold the maturity and duration mismatch in the meantime.) Selling these assets quickly means taking (large) haircuts on book value, meaning a random run on a healthy bank can destroy projects or at least turn the bank itself bankrupt -- destroying customer relatioships and disrupting the bank-money-level money supply etc etc.
That's all unnecessary, and a benevolent central bank/observer can improve upon the state of the world by injecting more money and support the healthy bank.
...except, depositors don't run on healthy banks!
That's what this beautiful history and database shows. Somehow, depositors (in aggregate) are clever enough to sniff out when something is wrong and not structurally/mindlessly/sunspot-style make erratic and concentrated mistakes.
If you scratch the lender or last resort literature a little, you find that it's a good, disciplining market mechanism for depositors to run on banks. Here's a schematic:
- if bank is healthy and other bankers know it, then the bank can source enough liquidity from them ("recycling" deposits), against a very minimal short-term expense. No problem, no negative outcome, no need for lender of last resort
- if the bank is unhealthy and the other bankers know it, they won't lend to it and it will fail, and it will be a good thing! Bankers taking bad risks or making bad investments should fail, like every other business in a market economy.
- if the bank is healthy, but the other bankers don't know it and so refuse to lend, that's when a lender of last resort needs to step in and save the day. By supplying that extra liquidity it can stave off the damaging run and everyone is better off for it.
Here's the best short article/overview I've ever seen on this (by George Kaufman, for econlib):
The literature on bank runs and central banks is basically an exercise in showing whether or not the central bank knows about the true state of a given bank better than other bankers do.
When central banks also had prudential obligation, by chance some of its workers would have recently audited and inspected private information for that bank, meaning it's quite plausible that central bankers do have some insider information here that they can act on. (Nowadays that function is usually separated to another agency, thus the argument is practically quite moot.)
But... If historically and practically depositors are good enough in aggregate to sniff out what's wrong, there's no need for a central bank to intervene and override their market choice.
Bank runs are bad for local economies
...only if the banks that are being abandoned have been making bad loans, a big borrower defaults, or something like that.
...in which case there is a real economic loss that we want/need someone to carry — not paper over and pretend it's not there. (_cough_, Bitfinex...)
Told yah:
Surely the mere process of a run on a healthy bank wreaked havoc on their local economy? Apparently not.
Cross-referencing each bank’s financial strength, the researchers trace the local economic impact of each bank run using Bradstreet’s contemporary reporting. They find that a run on a strong bank doesn’t really impose any economic cost at all on the local economy. Yet runs on banks whose fundamentals are awful are followed by major lending contractions that can cause all sorts of spillover problems.
and:
Out of 3,984 bank runs on individual US banks from 1863–1934, the authors identify 53 that look like they’re based on “misinformation” or “depositor confusion”, which reads as academese for “blind panic”. Of these 53 oddities, six banks failed, which sounds like maybe six too many. But the researchers argue that the historical data indicates that “strong banks (fundamentals in the upper tercile) subject to non-fundamental runs have a zero probability of failure”
one of the footnotes indicates that at least one of these failed banks had basically committed massive fraud, which was only uncovered by their so-called irrational run.
Even that, the depositors sniffed out. Amazing.
All "pretty interesting" says Nangle, understating the explosive result.
the ideas that the Nobel winners formalise are really the ones sitting behind the rationale for bank deposit insurance.
Yes. This evidence shows/weighs heavy that lender of last resort is a defunct and unnecessary function — liable to abuse, nepotism, and trigger-happy rulers, not actually saving a banking system that needs it
Their surprise about this is surprising. I remember a run on a south african mutual bank, where it later also turned that they were comitting massive fraud with government involvement. Which right now reminded of https://en.wikipedia.org/wiki/VBS_Mutual_Bank again.
Another higher order reason for why a run on a healthy bank wouldn't be too damaging is that, even if it causes that bank to fire-sale its assets, someone else is purchasing those assets. In many cases, those will be other banks that also correctly evaluate each specific asset as profitable.
So, even if other banks don't evaluate the bank being run on as being solvent over the entirety of its holdings, some banks should properly evaluate each separate holding and relatively few viable projects will actually shutdown.
I know some Austrians made this point back in the day (I wanna say Philip Bagus?), probs at a Mises Uni event, and it completely shocked me. Yes, of course the real world goes on REGARDLESS of what the financiers are up to!
Good point. Fire sale of assets only hurts if balance sheet was fake to begin with. Healthy bank just sells at market. Again, depositors sniff insolvency.
You're supposed to withdraw before it's very close to insolvent.
This part is powerful — depositors are actually good at telling which bank is dodgy, runs happen on already insolvent banks. Not panic, but pattern recognition.
We saw same in Nigeria 2023. When CBN did naira redesign, some banks had no cash but still showed big profits on paper. People queued at banks they knew were weak. The banks that managed cash well had no runs.
So if FT now says "runs kill solvent banks is overstated" — it means central banks' whole excuse for bailouts collapses. They don't protect solvent banks, they protect insolvent friends. Makes you understand why self-custody matters. Do you think that map 1800-1963 would look similar for Nigeria if someone scanned our newspapers with LLMs too?
Noted