Yes, but it’s pretty much always been problematic when a bank leader has had to make a statement akin to “We’re fine as long as there’s not a run”.

That’s the kind of thing that only gets said when there’s some concern that there will be a run.

They’re selling equity to get capital. That’s pretty dire straits, FTX was doing that before they went under (I’m not saying this is FTX, I’m just saying it can be akin to the nuclear option)

FTX's actions have nothing to do with what is happening with SVB, it's not even close. Why make a bad parallel?

The parallel is selling equity when you are in trouble is not ideal. Different reasons or sectors or whatever...but the principal is the same.

I mean it’s just as bad as buying it at the peak.

Their emergency stock sale was a key sign of their implosion. Healthy financial institutions never do this

For what reason besides raising money does a company ever sell shares in itself?

How does your statement make any sense?

Because in this case it’s a bank and they aren’t selling equity to invest in new business/products.

SVB had to liquidate good assets quickly, so it sold them at a loss. That means its capitalization compared to liabilities (deposits) could be in “uh oh” territory.

So it sells capital on the open market to shore up its liquid capital. Yes, most every capital raise works mostly like this. But it’s notable here because of what it portends.

> For what reason besides raising cash does a company ever sell shares in itself?

A healthy company doesn't generally sell shares in itself (except perhaps to fund a major expansion). Selling equity is a last resort when you don't have better funding options (retained earnings, debt, ...).

except when all of banks failed their stress tests, a) who remembers, b) who cared when it was announced?

All the banks passed their stress test this year

https://www.federalreserve.gov/newsevents/pressreleases/bcre...

Regulators have been putting immense pressure on big banks to hold adequate capital reserves since 2008 and they have been especially turning up the heat for the last 5 years. Moreover, it is clear that regulators will never allow a US bank to hold more than 3% of assets as crypto ever again.

Those government-mandated, ultra-safe capital reserves look like they're actually the big problem that's going to bring down banks right now. Banks have stuck a bunch of their reserves in really safe, predictable, high quality long-term bonds (particularly government issued ones). Because interest rates have gone up, those bonds are now worth substantially less than they were a year or so ago, meaning that the banks' ability to cope with people and businesses withdrawing their deposits has gone down substantially.

Interest on loans should by increase a banks reserves every year barring massive defaults. The ROI for the actual reserves aren’t particularly relevant by comparison.

Similarly from a reserve standpoint they don’t need to worry about inflation as they need to pay back deposits in nominal terms not what the money is worth when withdrawn.

Higher interest rates decrease the value of long-duration bonds even in nominal terms. Think of it this way: because the interest rate on a bond is fixed at issue, and because newer bonds now have higher interest rates, your existing bonds have to be sold for less in order for someone to buy them over a newer higher-interest bond. This means that the banks' reserves have actually shrunk in value. This is made worse by the fact that the combination of low interest rates and requirements to choose safe investments means that the only way for banks to meet their cost of operations is by picking longer-duration bonds which have higher drops in value, and that they haven't been earning enough interest to really grow their reserves either.

That’s only relevant if you need to sell it or use mark to market accounting.

The US banking system has been given a great deal of regulatory leeway due to recent economic turbulence, including setting reserve requirements to 0%. So market value is only relevant if they need to sell before maturity.

> So market value is only relevant if they need to sell before maturity.

Which is exactly what needs to happen when depositors ask for their money back.

Not typically.

Banks generally have liquid reserves to handle significant fluctuations in deposits. If that’s insufficient they have incoming cash flow and the option to borrow money to make up the difference rather than instantly selling assets.

Thus in practice extracting 5% over a week is fine but there’s a threshold that will kill any bank.

> Similarly from a reserve standpoint they don’t need to worry about inflation as they need to pay back deposits in nominal terms not what the money is worth when withdrawn.

The issue is that the sale value of their reserves has dropped below that nominal value. If you take in $1000 of deposits that you're paying 1% interest on and your reserve against that is a 10-year $1000 T-bill with a 2% coupon, you'd think you're fine, right? But if interest rates go up to 3%, you can't sell that T-bill for $1000 any more; if you can hold it to maturity you're fine, but if your customers start pulling their deposits you're in trouble.

Why would customers be pulling deposits unless you are offering lower than market interest rate? If T-bills are 3%, they can pay depositors 2% now and so whatever condition kept the customers there at -1% risk premium would still keep them there. No run on the bank. And given they are T-bills, duration is minimal, so $1000 might be worth $990 even before coupons. Whoop-de-doo!

There would only be a problem if the bank's credit deteriorated in its risky assets, enough to freak depositors out.

> Why would customers be pulling deposits unless you are offering lower than market interest rate?

Any number of reasons, particularly if all your customers are in the same industry. If you're "Silicon Valley Bank" and there's a downturn in Silicon Valley, well, here you are.

Yeah, this is how it's similar to silvergate. The main issue silvergate had was a lack of diversification of their liabilities, I.e. their depositors were all crypto. So when crypto hit a liquidity crises, it cascaded to silvergate. The same thing could theoretically happen to SVB, although that would look very bad on tech, since tech shouldn't really be an industry unto itself. Ideally, tech is just technology, and its application is across any industry.

This is incorrect, as we learned from Silvergate. Customers pulled their deposits for reasons unrelated to the bank (in Silvergate’s case, the customers of the crypto exchanges wanted all their money back, so the crypto exchanges had to withdraw their deposits from Silvergate).

As Levine put it, it’s not an asset problem, it’s a liability problem.

It is not unrelated to the bank. They took concentrated callable funding that somehow was not matched to the liquidity of their assets. They took a risk.

This doesn't work either. Remember, the reason that the market value of the T-bills in their reserves has dropped is that the interest rate on them is lower than the current interest rate that people can get by buying them now, so the bank likely cannot afford to pay their customers a market interest rate of 1% below that. The only way for the hold-to-maturity value of their reserves to actually be realised is if customers don't withdraw their deposits and they don't have to substantially increase the interest rates they pay in order to achieve that - otherwise they're likely to run out of money, either quickly or slowly.

You nailed it I think.

Right now there will be a lot of chatter if the root cause was industry concentration, mark to market, junior VCs scaring gulible founders, or as you pointed out: duration risk.

I think in five years the common narrative will be about duration risk.

To spell out what you hinted at: SVB will collapse at some point in the next few years. The only scenario they survive is if they survive this bank run, and then soonsih the fed lowers interest rates by a ton.

Otherwise even without this bank run SVB will still slowly be forced to sell off more and more HTM assets. Core reason being they cannot cashflow wise offer market rate interest on deposits. Clients _would_ move their money to better paying banks, not everyone but too many.

Thus this bank run only accelerates the inevitable. The interest broader note is how other bigger banks have or have not managed their exposure to duration. And special eyes towards the Japanese MegaBanks, double so if the bank of Japan does raise rates.

I was talking about offering it to new deposits. If your old deposits are not duration-matched then yes, you are screwed.

This is why there is a difference between a banks assets and their liquid assets.

If the T-Bill has a maturity beyond 90 days it doesn’t qualify as a liquid asset.

The parent commenter is right, and the problem is the mark-to-market rule. This means that the value of the asset must be the current trading value, which goes down as the rates go up. The result is that bank reserves will go down substantially in nominal terms, sometimes faster than they can recoup the value of these investments.

That’s a regulatory rule which can be suspended not the underlying economic reality. “On April 9, 2009, FASB issued an official update to FAS 157[35] that eases the mark-to-market rules when the market is unsteady or inactive.” https://en.wikipedia.org/wiki/Mark-to-market_accounting

We’re currently in some interesting times: “As announced on March 15, 2020, the Board reduced reserve requirement ratios to zero percent effective March 26, 2020. This action eliminated reserve requirements for all depository institutions.“. https://www.federalreserve.gov/monetarypolicy/reservereq.htm

The key phrase here is "when the market is unsteady or inactive". Apparently this can be used to save banks, but not until there is trouble in the market.

respectfully, I'm not so sure. The decline in bonds applies to all fixed-rate securities. The only alternatives would have been just straight up cash (bad with inflation) or riskier, less-liquid assets (non-tradable loans with floating rates, for example). They are limited on the latter by risk weighting, and I'm not sure having looser risk controls on the asset side would really help confidence in the banking sector.

please feel free to disagree!

Inflation isn’t a concern here. Depositors hand a bank 1 billion and X% inflation hits, well the bank still only owes them 1 billion.

Yeah that's true, it matter more for net-interest-margin / interest income generation in a rising-rate environment. Good point

a depositor isn't going to deposit 1 billion and only expect 1 billion back out months later. that's some poor people shit. Large sums of money like that, banks pay you for the pleasure of holding it. So a bank offering 4.5% APY like SVB offers would owe an extra 3.2 million to the billionaire if they look at their account a month later.

The problem is, the underlying assets that SVB owns will only pay them back at 1% APY, and only in 20 years or whatever, and the billionaire has been promised 4.5%APY and is expecting to have access to one months worth of interest next month. that 3.5 difference is thus a huge problem for SVB.

Could they instead hold short-term treasuries (as short as 4 weeks, I believe) and refuse to honor large withdraws until they mature?

Short term treasuries are definitely pretty common on the asset side, but if you refuse to honor withdrawals on demand deposits you won't have a bank anymore and the FDIC will step in to wind things down

Pardon my ignorance (not a crypto guy) — what is it that you're referencing when you mention a bank holding 3% of assets as crypto? My uninformed assumption is that this is either about Silvergate, or something related to SBF's debacle.

Just curious. As mentioned, I'm pretty ignorant about this and would like to educate myself.

As part of a general crackdown happening right now regulators put any crypto plans US banks may have had on ice.

We're fine as long as we don't hit an iceberg.

The first banks just stored your money and charge you for it, until somebody had a brilliant idea.

The very earliest banks didn't do fractional reserve banking so it isn't true to say it is how every bank worked since banks were invented. The first known instance of fractional reserve lending was with an medieval Italy but banks have been around since 2000BCE.

You are right that it definitely explains 99.99% of banks in history. But there are a handful of historical examples -- even after "modern banking" began in medieval Italy -- that don't fit. A recent example is The Narrow Bank which was shutdown by the Federal Reserve. But in the past you had things like the Bank of England in 1844 which went to 100% reserves for a period. Or banks under the Louisiana Banking Act of 1842 -- which was why banks in Louisiana were unaffected by the financial crisis of 1857.

So did Louisiana banks issue loans?

How could they issue loans? 10 x 10$ deposits means you can loan 100$? Where as the modern way is more like 100$ in deposits means you can lend out 1000$ because chances are everyone won’t not pay it back? And then can’t you say that since you’ve lent out 1000$ and chances are you’ll get paid back, you’ve basically got 1104.56$ and so can lend out 10k$? And then you bundle those together and sell them to each other depending on what ratio of income to cash you want?

Apologies if no one was meant to answer that.

Doesn't stop a run

10 people put $10 in your bank. You give someone a loan for $50 and leave $50 in the vault. 7 of your customers take $10 out, you are screwed.

Depends. If they take it out electrically, send it to another bank then it just shows up as debt in a database from bank A to bank B. Cash is dirty and boring nowadays.

This doesn't apply to certificates of deposit as depositors agree to not withdraw their money until a specified point in time.

Especially in contemporary times banks make money in an immense amount of ways that don't involve touching customer funds: debit transaction fees, international exchange rate "adjustments", ATM fees, the million 'special processing fee' type fees, and so on. In other countries I've even had to pay a fee when depositing, which was quite odd.

Of course this all is going to pale in comparison to the amount that banks make by d̶u̶m̶p̶i̶n̶g̶ ̶c̶o̶n̶s̶u̶m̶e̶r̶ ̶f̶u̶n̶d̶s̶,̶ ̶h̶e̶a̶v̶i̶l̶y̶ ̶l̶e̶v̶e̶r̶a̶g̶e̶d̶,̶ ̶i̶n̶t̶o̶ ̶h̶i̶g̶h̶ ̶r̶i̶s̶k̶ ̶a̶s̶s̶e̶t̶s̶ responsibly investing deposits. But of course banks under '100% deposits maintained' type systems could then engage in more typical behavior with their own funds above and beyond what's made from deposits. Under such a regime no bank would ever be "too big to fail", customer deposits would be 100% guaranteed at all times, and more. In exchange you'd see substantially slower overall economic growth and monetary multiplication, but I'm increasingly convinced that would not have been a bad thing.

> In other countries I've even had to pay a fee when depositing, which was quite odd.

That's normal for business banking I think. Trucking cash and coins around isn't free, that stuff is heavy.

The Medicis couldn’t loan money at interest— that was usury—so they made money by charging a fee for allowing customers to deposit at one place and have the money available in other locations. That was not considered usury

It’s amazing how long this goes back. ATM transactions fees are stupid high, and now I have someone to blame.

I don't believe the claim that non-fractional reserve banking would actually slow economic growth.

Is real economic growth even determined by anything but technological development?

Of course, the economy can be made to "grow" by some slight of hand, like having a high inflation rate while pretending that we don't. Or by depleting natural resources. But that's not the kind of growth we want.

Some loans go to businesses so they can buy a new widget-making machine, employ more operators, and profitably sell widgets. Economic growth in action!

Other loans go to people who were going to buy a doodad after saving up for 12 months, who instead get the doodad immediately and pay for it for 14 months. That looks like economic growth, because in month 1 doodad sales have risen. But if the sale would have happened anyway, the 'growth' is lot more debatable IMHO.

So if no one got those loans, what would have happened to the resources they bought? What would've happened to the resources that went into building the widget making machine? What would have happened to the operator employees? Would they've disappeared? Most likely they would've become available to another investor who did not get the loan but also wanted to build a widget making machine.

Did the loan actually increase economic growth? I think the only reasonable answer is: Yes, if the bank issuing the loan had a better idea than the market about the future profitability of the investment. However, that doesn't seem very likely to me.

>Is real economic growth even determined by anything but technological development?

That is just the end result, the question is how do you get there? How do you organize an economy to reach that outcome in the most optimal way?

Okay, I agree, that's the question. So how does fractional reserve banking contribute to that organization?

It helps if you take a deep breath.

> Where as the modern way is more like 100$ in deposits means you can lend out 1000$

More like 20,000$, but yes.

Yes, they still issued loans. Louisiana banks weren't true "narrow banks" but were much tighter regulated with much higher reserve requirements than other states at the time (or today). This was a reaction to the financial crisis of 1837.

> The very earliest banks didn't do fractional reserve banking

I am certain they all did it, but in secrecy. Even when banks stocked grains and not gold.